id
string | submitter
string | authors
string | title
string | comments
string | journal-ref
string | doi
string | report-no
string | categories
string | license
string | abstract
string | versions
list | update_date
timestamp[s] | authors_parsed
sequence |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2505.21878 | Mingjian Cheng | Wenjie Jiang, Mingjian Cheng, Lixin Guo, Xiang Yi, Jiangting Li, Junli Wang and Andrew Forbes | Machine learning assisted speckle and OAM spectrum analysis for enhanced turbulence characterisation | null | null | null | null | physics.optics | http://creativecommons.org/licenses/by-sa/4.0/ | Atmospheric turbulence presents a major obstacle to the performance of free-space optical (FSO) communication and environmental sensing. While most studies focused on robust detection and recognition of transmitted structured light in turbulent media, a critical, yet often underexplored, avenue involves leveraging the structure of light itself to detect and characterize the medium itself. Here we introduce a deep learning framework that synergistically leverages post-transmission intensity speckle patterns and orbital angular momentum (OAM) spectral data to enhance turbulence state classification, essential for applications such as energy delivery, effective adaptive optics and communications. The proposed architecture builds upon an enhanced InceptionNet backbone optimized for multi-scale feature extraction from complex optical inputs, where we demonstrate that multiple degree of freedom analysis outperforms the conventional single parameter approaches, reaching validation accuracies in excess of 80%. Our approach is benchmarked against standard implementations and found to be superior in handling diverse turbulence scenarios, demonstrating enhanced stability across varying turbulence conditions defined by different Reynolds numbers and Fried parameters. Our method offers a scalable, data-efficient solution for precise atmospheric turbulence detection, with strong potential for deployment in real-world optical sensing applications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 01:48:12 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Jiang",
"Wenjie",
""
],
[
"Cheng",
"Mingjian",
""
],
[
"Guo",
"Lixin",
""
],
[
"Yi",
"Xiang",
""
],
[
"Li",
"Jiangting",
""
],
[
"Wang",
"Junli",
""
],
[
"Forbes",
"Andrew",
""
]
] |
2505.21879 | Jiaxu Cui | Weiting Liu and Jiaxu Cui and Jiao Hu and En Wang and Bo Yang | Symbolic Foundation Regressor on Complex Networks | 60 pages | null | null | null | cs.SC cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | In science, we are interested not only in forecasting but also in understanding how predictions are made, specifically what the interpretable underlying model looks like. Data-driven machine learning technology can significantly streamline the complex and time-consuming traditional manual process of discovering scientific laws, helping us gain insights into fundamental issues in modern science. In this work, we introduce a pre-trained symbolic foundation regressor that can effectively compress complex data with numerous interacting variables while producing interpretable physical representations. Our model has been rigorously tested on non-network symbolic regression, symbolic regression on complex networks, and the inference of network dynamics across various domains, including physics, biochemistry, ecology, and epidemiology. The results indicate a remarkable improvement in equation inference efficiency, being three times more effective than baseline approaches while maintaining accurate predictions. Furthermore, we apply our model to uncover more intuitive laws of interaction transmission from global epidemic outbreak data, achieving optimal data fitting. This model extends the application boundary of pre-trained symbolic regression models to complex networks, and we believe it provides a foundational solution for revealing the hidden mechanisms behind changes in complex phenomena, enhancing interpretability, and inspiring further scientific discoveries. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 01:53:29 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Liu",
"Weiting",
""
],
[
"Cui",
"Jiaxu",
""
],
[
"Hu",
"Jiao",
""
],
[
"Wang",
"En",
""
],
[
"Yang",
"Bo",
""
]
] |
2505.21880 | Yu-Ming Chang | Yu-Lun Song, Chung-En Tsern, Che-Cheng Wu, Yu-Ming Chang, Syuan-Bo Huang, Wei-Chu Chen, Michael Chia-Liang Lin, Yu-Ta Lin | Incorporating LLMs for Large-Scale Urban Complex Mobility Simulation | 8 pages, 8 figures. This paper is reviewed and accepted by the CUPUM (Computational Urban Planning and Urban Management) Conference held by University College London (UCL) in 2025 | null | null | null | cs.MA cs.AI cs.CL cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study presents an innovative approach to urban mobility simulation by integrating a Large Language Model (LLM) with Agent-Based Modeling (ABM). Unlike traditional rule-based ABM, the proposed framework leverages LLM to enhance agent diversity and realism by generating synthetic population profiles, allocating routine and occasional locations, and simulating personalized routes. Using real-world data, the simulation models individual behaviors and large-scale mobility patterns in Taipei City. Key insights, such as route heat maps and mode-specific indicators, provide urban planners with actionable information for policy-making. Future work focuses on establishing robust validation frameworks to ensure accuracy and reliability in urban planning applications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 01:54:28 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Song",
"Yu-Lun",
""
],
[
"Tsern",
"Chung-En",
""
],
[
"Wu",
"Che-Cheng",
""
],
[
"Chang",
"Yu-Ming",
""
],
[
"Huang",
"Syuan-Bo",
""
],
[
"Chen",
"Wei-Chu",
""
],
[
"Lin",
"Michael Chia-Liang",
""
],
[
"Lin",
"Yu-Ta",
""
]
] |
2505.21881 | Junjie Liu | Rui Guan and Junjie Liu and Jian-Hua Jiang | Global-Local Duality of Energetic Control Cost in Multipartite Quantum Correlated Systems | Accepted by Physical Review Research, in press | null | null | null | quant-ph cond-mat.stat-mech | http://creativecommons.org/licenses/by/4.0/ | Multipartite quantum correlated systems (MQCSs) are widely utilized in diverse quantum information tasks, where their sophisticated control inherently incurs energetic costs. However, the fundamental characteristics of these control costs remain elusive, largely due to the lack of thermodynamic descriptions capable of capturing the full complexities of MQCSs. Here, we uncover universal thermodynamic relations for arbitrary MQCSs weakly coupled to a thermal bath, establishing an intrinsic global-local duality of control costs. Using these relations, we elucidate the exact role of multipartite correlation--a defining quantum feature of MQCSs--in shaping control costs at finite times. We also demonstrate that the relative magnitude between global and local control costs is undetermined, which perplexes the cost management of MQCSs under finite-time controls. Our results are numerically corroborated with applications to experimentally realizable multi-qubit systems undergoing finite-time qubit reset processes. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 01:56:12 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Guan",
"Rui",
""
],
[
"Liu",
"Junjie",
""
],
[
"Jiang",
"Jian-Hua",
""
]
] |
2505.21883 | Shu Lin | Si-wen Li and Shu Lin | Spin polarization of holographic baryon in strongly coupled fluid | 28 pages, 4 figures | null | null | null | hep-th hep-ph nucl-th | http://creativecommons.org/licenses/by/4.0/ | The spin polarization for baryon in a hydrodynamic medium has been extensively studied in the weakly coupled regime using quantum kinetic theory. As a first study of this problem in the strongly coupled regime, we investigate holographically the spectral function of a probe baryon in the fluid-gravity background. This is done by carefully performing gradient expansion of the Dirac equation in the fluid-gravity background. Different contributions in the expansion are understood in terms of density matrices of the probe baryon and the medium. The resulting spectral functions indicate that the holographic baryonic is polarized as responses to fluid acceleration, shear stress and vorticity. The structures of the responses are similar to those found in weakly coupled studies. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 01:59:47 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Li",
"Si-wen",
""
],
[
"Lin",
"Shu",
""
]
] |
2505.21884 | Junhong Liu Mr. | Junhong Liu, Qinfei Long, Rong-Peng Liu, Wenjie Liu, Yunhe Hou | Online distributed optimization for spatio-temporally constrained real-time peer-to-peer energy trading | null | Applied Energy 331 (2023): 120216 | 10.1016/j.apenergy.2022.120216 | null | eess.SY cs.SY math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The proliferation of distributed renewable energy triggers the peer-to-peer (P2P) energy market formations. To make profits, prosumers equipped with photovoltaic (PV) panels and even the energy storage system (ESS) can actively participate in the real-time P2P energy market and trade energy. However, in real situations, system states such as energy demands and renewable energy power generation are highly uncertain, making it difficult for prosumers to make optimal real-time decisions. Moreover, severe problems with the physical network can arise from the real-time P2P energy trading, such as bus voltage violations and line overload. To handle these problems, this work first formulates the real-time P2P energy trading problem as a spatio-temporally constrained stochastic optimization problem by considering ESS and the spatial physical network constraints. To deal with the uncertainties online, a modified Lyapunov optimization method is innovatively proposed to approximately reformulate the stochastic optimization problem into an online one by relaxing the time-coupling constraints. Compared with the state-of-the-art online methods, the proposed one renders more flexibility and better performance for the real-time P2P energy market operation. Additionally, to protect the prosumers' privacy, an online distributed algorithm based on the consensus alternating direction method of multipliers (ADMM) is developed to solve the reformulated online problem by decoupling the spatial constraints. The theoretical near-optimal performance guarantee of the proposed online distributed algorithm is derived, and its performance can be further improved by minimizing the performance gap. Simulation results demonstrate that the proposed online distributed algorithm can guarantee the fast, stable, and safe long-term operation of the real-time P2P energy market. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 01:59:59 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Liu",
"Junhong",
""
],
[
"Long",
"Qinfei",
""
],
[
"Liu",
"Rong-Peng",
""
],
[
"Liu",
"Wenjie",
""
],
[
"Hou",
"Yunhe",
""
]
] |
2505.21885 | James Dent | Rouzbeh Allahverdi, James B. Dent, Ngo Phuc Duc Loc, Tao Xu | Gravitational wave signatures of primordial black hole accretion during early matter domination | 14 pages + 2 pages of Appendices | null | null | CETUP-2024-011 | hep-ph astro-ph.CO gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a scenario in which primordial black holes (PBHs) form in a post-inflationary radiation-dominated (RD) phase and then experience significant accretion during a phase of early matter dominated (EMD). We show that PBH masses could grow by up to two orders of magnitude. Restricting to the linear perturbation regime, we compute the gravitational wave (GW) spectrum that features two peaks. The high-frequency peak is associated with the PBH formation in the RD phase, while the low-frequency peak is due to the sudden transition from EMD to the later, standard RD phase. We identify a PBH mass range where one or both peaks can be observed by a combination of different GW detectors. Finally, we show the signal-to-noise ratio of the total GW spectrum for PBHs in the asteroid mass window, where they could comprise the totality of dark matter. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:00:33 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Allahverdi",
"Rouzbeh",
""
],
[
"Dent",
"James B.",
""
],
[
"Loc",
"Ngo Phuc Duc",
""
],
[
"Xu",
"Tao",
""
]
] |
2505.21886 | Haotian Chen | Haotian Chen, Xishuo Wei, Hongxuan Zhu, and Zhihong Lin | Geometry effects on zonal flow dynamics and turbulent transport in optimized stellarators | 4 pages, 4 figures,1 table | null | null | null | physics.plasm-ph | http://creativecommons.org/licenses/by/4.0/ | Global gyrokinetic simulations find a strong suppression of ion temperature gradient (ITG) turbulence by zonal flows in stellarators optimized for neoclassical transport. The reduction of the ITG transport by the zonal flows in quasi-helicalsymmetric (QH) and quasi-isodynamic (QI) stellarators are much larger than a quasi-axisymmetric (QA) stellarator or a tokamak, thanks to higher linear residual levels and lower nonlinear frequencies of the zonal flows in the QH and QI. The transport level and energy confinement time in the QH and QI are similar to the tokamak with the same size and temperature gradient, despite the much larger linear growth rates in the stellarators. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:01:19 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Chen",
"Haotian",
""
],
[
"Wei",
"Xishuo",
""
],
[
"Zhu",
"Hongxuan",
""
],
[
"Lin",
"Zhihong",
""
]
] |
2505.21888 | Hsing-Ta Chen | Guangming Liu, Hsing-Ta Chen | Dissecting Exciton-Polariton Transport in Organic Molecular Crystals: Emerging Conductivity Assisted by Intermolecular Vibrational Coupling | null | null | null | null | physics.chem-ph cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | In this work, we systematically investigate the spectral and transport properties of exciton-polaritons under the explicit influence of intermolecular vibrational coupling, which introduces dynamic disorder. In the context of a one-dimensional molecular chain strongly interacting with a cavity photon, we demonstrate the polaritonic characteristics of the spectral function and its interactions with the electronic band broadened by the coupling disorder. We further dissect the current flux into its bare excitonic contribution and transport via the cavity photon. Our results reveal that the enhancement in the charge carrier mobility and frequency-resolved conductivity stems from the photon-mediated current. More importantly, contrary to the intuition that dynamic disorder hinders transport, intermolecular vibrational coupling can facilitate exciton-polariton transport, offering an additional degree of tunability for material properties. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:03:57 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Liu",
"Guangming",
""
],
[
"Chen",
"Hsing-Ta",
""
]
] |
2505.21890 | Sunil Kumar Narayanan | Sunil Kumar Narayanan, Lingjun Zhao, Lu Gan, Yongsheng Chen | Hyperspectral Gaussian Splatting | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Hyperspectral imaging (HSI) has been widely used in agricultural applications for non-destructive estimation of plant nutrient composition and precise determination of nutritional elements in samples. Recently, 3D reconstruction methods have been used to create implicit neural representations of HSI scenes, which can help localize the target object's nutrient composition spatially and spectrally. Neural Radiance Field (NeRF) is a cutting-edge implicit representation that can render hyperspectral channel compositions of each spatial location from any viewing direction. However, it faces limitations in training time and rendering speed. In this paper, we propose Hyperspectral Gaussian Splatting (HS-GS), which combines the state-of-the-art 3D Gaussian Splatting (3DGS) with a diffusion model to enable 3D explicit reconstruction of the hyperspectral scenes and novel view synthesis for the entire spectral range. To enhance the model's ability to capture fine-grained reflectance variations across the light spectrum and leverage correlations between adjacent wavelengths for denoising, we introduce a wavelength encoder to generate wavelength-specific spherical harmonics offsets. We also introduce a novel Kullback--Leibler divergence-based loss to mitigate the spectral distribution gap between the rendered image and the ground truth. A diffusion model is further applied for denoising the rendered images and generating photorealistic hyperspectral images. We present extensive evaluations on five diverse hyperspectral scenes from the Hyper-NeRF dataset to show the effectiveness of our proposed HS-GS framework. The results demonstrate that HS-GS achieves new state-of-the-art performance among all previously published methods. Code will be released upon publication. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:07:52 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Narayanan",
"Sunil Kumar",
""
],
[
"Zhao",
"Lingjun",
""
],
[
"Gan",
"Lu",
""
],
[
"Chen",
"Yongsheng",
""
]
] |
2505.21891 | Yun Suen Pai | Arooj Zaidi, Giulia Barbareschi, Kai Kunze, Yun Suen Pai, Junichi Yamaoka | TIEboard: A Digital Educational Tool for Kids Geometric Learning | null | Proc. ACM Interactive Mobile Wearable Ubiquitous Technology, Vol. 9, No. 2, Article 62. Publication date: June 2025 | 10.1145/3729478 | null | cs.HC | http://creativecommons.org/licenses/by/4.0/ | Tangible User Interfaces have shown potential in supporting the acquisition of key concepts in computing and mathematics while fostering engagement in young learners, but these approaches are less commonly utilised in the context of geometry. In this paper we introduce TIEboard, an interactive device to promote early learning of basic geometry concepts. TIEboard draws inspiration from traditional geoboards and lacing toys to leverage children's familiarity with these traditional tools. It employs instructional lights to guide children in creating shapes using colourful threads of optical fiber. The use of conductive materials allows the system to detect lacing activity and provide feedback in real-time. TIEboard incorporates six interaction modes of varying difficulty based on an incremental learning framework. The study evaluated TIEboard's effectiveness in supporting early geometric learning, facilitating creativity and promoting collaboration among 16 children aged 5-9. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:10:03 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zaidi",
"Arooj",
""
],
[
"Barbareschi",
"Giulia",
""
],
[
"Kunze",
"Kai",
""
],
[
"Pai",
"Yun Suen",
""
],
[
"Yamaoka",
"Junichi",
""
]
] |
2505.21892 | Xunpeng Huang | Xunpeng Huang, Yingyu Lin, Nikki Lijing Kuang, Hanze Dong, Difan Zou, Yian Ma, Tong Zhang | Almost Linear Convergence under Minimal Score Assumptions: Quantized Transition Diffusion | 37 pages, 3 figures, 3 tables | null | null | null | stat.ML cs.LG | http://creativecommons.org/licenses/by/4.0/ | Continuous diffusion models have demonstrated remarkable performance in data generation across various domains, yet their efficiency remains constrained by two critical limitations: (1) the local adjacency structure of the forward Markov process, which restricts long-range transitions in the data space, and (2) inherent biases introduced during the simulation of time-inhomogeneous reverse denoising processes. To address these challenges, we propose Quantized Transition Diffusion (QTD), a novel approach that integrates data quantization with discrete diffusion dynamics. Our method first transforms the continuous data distribution $p_*$ into a discrete one $q_*$ via histogram approximation and binary encoding, enabling efficient representation in a structured discrete latent space. We then design a continuous-time Markov chain (CTMC) with Hamming distance-based transitions as the forward process, which inherently supports long-range movements in the original data space. For reverse-time sampling, we introduce a \textit{truncated uniformization} technique to simulate the reverse CTMC, which can provably provide unbiased generation from $q_*$ under minimal score assumptions. Through a novel KL dynamic analysis of the reverse CTMC, we prove that QTD can generate samples with $O(d\ln^2(d/\epsilon))$ score evaluations in expectation to approximate the $d$--dimensional target distribution $p_*$ within an $\epsilon$ error tolerance. Our method not only establishes state-of-the-art inference efficiency but also advances the theoretical foundations of diffusion-based generative modeling by unifying discrete and continuous diffusion paradigms. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:10:11 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Huang",
"Xunpeng",
""
],
[
"Lin",
"Yingyu",
""
],
[
"Kuang",
"Nikki Lijing",
""
],
[
"Dong",
"Hanze",
""
],
[
"Zou",
"Difan",
""
],
[
"Ma",
"Yian",
""
],
[
"Zhang",
"Tong",
""
]
] |
2505.21893 | Xiaomeng Yang | Xiaomeng Yang, Zhiyu Tan, Junyan Wang, Zhijian Zhou, Hao Li | SDPO: Importance-Sampled Direct Preference Optimization for Stable Diffusion Training | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Preference learning has become a central technique for aligning generative models with human expectations. Recently, it has been extended to diffusion models through methods like Direct Preference Optimization (DPO). However, existing approaches such as Diffusion-DPO suffer from two key challenges: timestep-dependent instability, caused by a mismatch between the reverse and forward diffusion processes and by high gradient variance in early noisy timesteps, and off-policy bias arising from the mismatch between optimization and data collection policies. We begin by analyzing the reverse diffusion trajectory and observe that instability primarily occurs at early timesteps with low importance weights. To address these issues, we first propose DPO-C\&M, a practical strategy that improves stability by clipping and masking uninformative timesteps while partially mitigating off-policy bias. Building on this, we introduce SDPO (Importance-Sampled Direct Preference Optimization), a principled framework that incorporates importance sampling into the objective to fully correct for off-policy bias and emphasize informative updates during the diffusion process. Experiments on CogVideoX-2B, CogVideoX-5B, and Wan2.1-1.3B demonstrate that both methods outperform standard Diffusion-DPO, with SDPO achieving superior VBench scores, human preference alignment, and training robustness. These results highlight the importance of timestep-aware, distribution-corrected optimization in diffusion-based preference learning. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:11:56 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Yang",
"Xiaomeng",
""
],
[
"Tan",
"Zhiyu",
""
],
[
"Wang",
"Junyan",
""
],
[
"Zhou",
"Zhijian",
""
],
[
"Li",
"Hao",
""
]
] |
2505.21894 | Yuanyuan Liu | Yuanyuan Liu, Yuanbiao Yang, Zhuo-Xu Cui, Qingyong Zhu, Jing Cheng, Congcong Liu, Jinwen Xie, Jingran Xu, Hairong Zheng, Dong Liang, Yanjie Zhu | Patch-based Reconstruction for Unsupervised Dynamic MRI using Learnable Tensor Function with Implicit Neural Representation | null | null | null | null | eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Dynamic MRI plays a vital role in clinical practice by capturing both spatial details and dynamic motion, but its high spatiotemporal resolution is often limited by long scan times. Deep learning (DL)-based methods have shown promising performance in accelerating dynamic MRI. However, most existing algorithms rely on large fully-sampled datasets for training, which are difficult to acquire. Recently, implicit neural representation (INR) has emerged as a powerful scan-specific paradigm for accelerated MRI, which models signals as a continuous function over spatiotemporal coordinates. Although this approach achieves efficient continuous modeling of dynamic images and robust reconstruction, it faces challenges in recovering fine details and increasing computational demands for high dimensional data representation. To enhance both efficiency and reconstruction quality, we propose TenF-INR, a novel patch-based unsupervised framework that employs INR to model bases of tensor decomposition, enabling efficient and accurate modeling of dynamic MR images with learnable tensor functions. By exploiting strong correlations in similar spatial image patches and in the temporal direction, TenF-INR enforces multidimensional low-rankness and implements patch-based reconstruction with the benefits of continuous modeling. We compare TenF-INR with state-of-the-art methods, including supervised DL methods and unsupervised approaches. Experimental results demonstrate that TenF-INR achieves high acceleration factors up to 21, outperforming all comparison methods in image quality, temporal fidelity, and quantitative metrics, even surpassing the supervised methods. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:13:09 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Liu",
"Yuanyuan",
""
],
[
"Yang",
"Yuanbiao",
""
],
[
"Cui",
"Zhuo-Xu",
""
],
[
"Zhu",
"Qingyong",
""
],
[
"Cheng",
"Jing",
""
],
[
"Liu",
"Congcong",
""
],
[
"Xie",
"Jinwen",
""
],
[
"Xu",
"Jingran",
""
],
[
"Zheng",
"Hairong",
""
],
[
"Liang",
"Dong",
""
],
[
"Zhu",
"Yanjie",
""
]
] |
2505.21895 | Cameron Gordon | Cameron Gordon, Yiping Ji, Hemanth Saratchandran, Paul Albert, Simon Lucey | Compressing Sine-Activated Low-Rank Adapters through Post-Training Quantization | 23 pages, 9 figures | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Low-Rank Adaptation (LoRA) has become a standard approach for parameter-efficient fine-tuning, offering substantial reductions in trainable parameters by modeling updates as the product of two low-rank matrices. While effective, the low-rank constraint inherently limits representational capacity, often resulting in reduced performance compared to full-rank fine-tuning. Recent work by Ji et al. (2025) has addressed this limitation by applying a fixed-frequency sinusoidal transformation to low-rank adapters, increasing their stable rank without introducing additional parameters. This raises a crucial question: can the same sine-activated technique be successfully applied within the context of Post-Training Quantization to retain benefits even after model compression? In this paper, we investigate this question by extending the sinusoidal transformation framework to quantized LoRA adapters. We develop a theoretical analysis showing that the stable rank of a quantized adapter is tightly linked to that of its full-precision counterpart, motivating the use of such rank-enhancing functions even under quantization. Our results demonstrate that the expressivity gains from a sinusoidal non-linearity persist after quantization, yielding highly compressed adapters with negligible loss in performance. We validate our approach across a range of fine-tuning tasks for language, vision and text-to-image generation achieving significant memory savings while maintaining competitive accuracy. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:15:15 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Gordon",
"Cameron",
""
],
[
"Ji",
"Yiping",
""
],
[
"Saratchandran",
"Hemanth",
""
],
[
"Albert",
"Paul",
""
],
[
"Lucey",
"Simon",
""
]
] |
2505.21896 | Reuben Wang | Reuben R. W. Wang and John L. Bohn | Theory of itinerant collisional spin dynamics in nondegenerate molecular gases | 22 pages, 9 figures | null | null | null | cond-mat.quant-gas cond-mat.dis-nn physics.atom-ph quant-ph | http://creativecommons.org/licenses/by/4.0/ | We study the fully itinerant dynamics of ultracold but nondegenerate polar molecules with a spin-$1/2$ degree of freedom encoded into two of their electric field dressed rotational states. Center of mass molecular motion is constrained to two-dimensions via tight confinement with a one-dimensional optical lattice, but remains mostly unconstrained within the plane. The pseudospins can become entangled through ultracold dipolar collisions, for which the locality of interactions is greatly relaxed by free molecular motion. At the level of single-molecule observables, collision-induced entanglement manifests as spin decoherence, for which our theoretical calculations serve well to describe recent Ramsey contrast measurements of quasi-2D confined KRb molecules at JILA [A. Carroll et al., Science 388 6745 (2025)]. In presenting a more detailed theoretical analysis of the KRb experiment, we highlight a key finding that molecular loss enhanced by particle exchange symmetry can lead to a suppression of collective spin decoherence, a mechanism with refer to as ``loss-induced quantum autoselection". We then show that by utilizing bialkali species with sufficiently large dipole moments, loss can be near completely suppressed in all collision channels via electric field tunable confinement-induced collisional shielding. The afforded collisional stability permits fully coherent spin mixing dynamics, natively realizing unitary circuit dynamics with random all-to-all connectivity and U(1) charge conservation. This work establishes a bridge between the domains of ultracold molecular collisions and many-body spin physics, ultimately proposing the use of nondegenerate bulk molecular gases as a controllable platform for nonequilibrium explorations of itinerant quantum matter. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:18:22 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wang",
"Reuben R. W.",
""
],
[
"Bohn",
"John L.",
""
]
] |
2505.21897 | Jianchao Jiang | Jianchao Jiang and Haofeng Zhang | Concentrate on Weakness: Mining Hard Prototypes for Few-Shot Medical Image Segmentation | 12 pages, 9 figures, 9 tables, accepted by IJCAI 2025 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Few-Shot Medical Image Segmentation (FSMIS) has been widely used to train a model that can perform segmentation from only a few annotated images. However, most existing prototype-based FSMIS methods generate multiple prototypes from the support image solely by random sampling or local averaging, which can cause particularly severe boundary blurring due to the tendency for normal features accounting for the majority of features of a specific category. Consequently, we propose to focus more attention to those weaker features that are crucial for clear segmentation boundary. Specifically, we design a Support Self-Prediction (SSP) module to identify such weak features by comparing true support mask with one predicted by global support prototype. Then, a Hard Prototypes Generation (HPG) module is employed to generate multiple hard prototypes based on these weak features. Subsequently, a Multiple Similarity Maps Fusion (MSMF) module is devised to generate final segmenting mask in a dual-path fashion to mitigate the imbalance between foreground and background in medical images. Furthermore, we introduce a boundary loss to further constraint the edge of segmentation. Extensive experiments on three publicly available medical image datasets demonstrate that our method achieves state-of-the-art performance. Code is available at https://github.com/jcjiang99/CoW. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:22:05 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Jiang",
"Jianchao",
""
],
[
"Zhang",
"Haofeng",
""
]
] |
2505.21898 | Rennai Qiu | Rennai Qiu, Chen Qian, Ran Li, Yufan Dang, Weize Chen, Cheng Yang, Yingli Zhang, Ye Tian, Xuantang Xiong, Lei Han, Zhiyuan Liu, Maosong Sun | Co-Saving: Resource Aware Multi-Agent Collaboration for Software Development | Work in Progress | null | null | null | cs.CL cs.AI cs.MA cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advancements in Large Language Models (LLMs) and autonomous agents have demonstrated remarkable capabilities across various domains. However, standalone agents frequently encounter limitations when handling complex tasks that demand extensive interactions and substantial computational resources. Although Multi-Agent Systems (MAS) alleviate some of these limitations through collaborative mechanisms like task decomposition, iterative communication, and role specialization, they typically remain resource-unaware, incurring significant inefficiencies due to high token consumption and excessive execution time. To address these limitations, we propose a resource-aware multi-agent system -- Co-Saving (meaning that multiple agents collaboratively engage in resource-saving activities), which leverages experiential knowledge to enhance operational efficiency and solution quality. Our key innovation is the introduction of "shortcuts" -- instructional transitions learned from historically successful trajectories -- which allows to bypass redundant reasoning agents and expedite the collective problem-solving process. Experiments for software development tasks demonstrate significant advantages over existing methods. Specifically, compared to the state-of-the-art MAS ChatDev, our method achieves an average reduction of 50.85% in token usage, and improves the overall code quality by 10.06%. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:23:53 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Qiu",
"Rennai",
""
],
[
"Qian",
"Chen",
""
],
[
"Li",
"Ran",
""
],
[
"Dang",
"Yufan",
""
],
[
"Chen",
"Weize",
""
],
[
"Yang",
"Cheng",
""
],
[
"Zhang",
"Yingli",
""
],
[
"Tian",
"Ye",
""
],
[
"Xiong",
"Xuantang",
""
],
[
"Han",
"Lei",
""
],
[
"Liu",
"Zhiyuan",
""
],
[
"Sun",
"Maosong",
""
]
] |
2505.21899 | Li Rui | Jianfei Liu, Rui Li, Zhilin Yang, Peichang Shi, Guodong Yi and Huaimin Wang | Joint$\lambda$: Orchestrating Serverless Workflows on Jointcloud FaaS Systems | null | null | null | null | cs.DC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Existing serverless workflow orchestration systems are predominantly designed for a single-cloud FaaS system, leading to vendor lock-in. This restricts performance optimization, cost reduction, and availability of applications. However, orchestrating serverless workflows on Jointcloud FaaS systems faces two main challenges: 1) Additional overhead caused by centralized cross-cloud orchestration; and 2) A lack of reliable failover and fault-tolerant mechanisms for cross-cloud serverless workflows. To address these challenges, we propose Joint$\lambda$, a distributed runtime system designed to orchestrate serverless workflows on multiple FaaS systems without relying on a centralized orchestrator. Joint$\lambda$ introduces a compatibility layer, Backend-Shim, leveraging inter-cloud heterogeneity to optimize makespan and reduce costs with on-demand billing. By using function-side orchestration instead of centralized nodes, it enables independent function invocations and data transfers, reducing cross-cloud communication overhead. For high availability, it ensures exactly-once execution via datastores and failover mechanisms for serverless workflows on Jointcloud FaaS systems. We validate Joint$\lambda$ on two heterogeneous FaaS systems, AWS and ALiYun, with four workflows. Compared to the most advanced commercial orchestration services for single-cloud serverless workflows, Joint$\lambda$ reduces up to 3.3$\times$ latency, saving up to 65\% cost. Joint$\lambda$ is also faster than the state-of-the-art orchestrators for cross-cloud serverless workflows up to 4.0$\times$, reducing up to 4.5$\times$ cost and providing strong execution guarantees. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:24:12 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Liu",
"Jianfei",
""
],
[
"Li",
"Rui",
""
],
[
"Yang",
"Zhilin",
""
],
[
"Shi",
"Peichang",
""
],
[
"Yi",
"Guodong",
""
],
[
"Wang",
"Huaimin",
""
]
] |
2505.21900 | Hyukpyo Hong | Hyukpyo Hong, Diego Rojas La Luz, Gheorghe Craciun | Ubiquitous Asymptotic Robustness in Biochemical Systems | This include two files: a main text and Supplementary Information. 17 pages, 4 figures, 2 tables for the main text; 19 pages, 1 figure, 18 tables for the Supplementary Information | null | null | null | math.DS q-bio.QM | http://creativecommons.org/licenses/by/4.0/ | Living systems maintain stable internal states despite environmental fluctuations. Absolute concentration robustness (ACR) is a striking homeostatic phenomenon in which the steady-state concentration of a molecular species remains entirely invariant to changes in total molecular supply. Although experimental studies have reported approximate -- but not exact -- robustness in steady-state concentrations, such behavior has often been attributed to exact ACR motifs perturbed by measurement noise or minor side reactions, rather than recognized as a structural property of the network itself. In this work, we highlight a previously underappreciated phenomenon, which we term {\em asymptotic ACR} (or {\em aACR}): approximate robustness can emerge solely from the architecture of the reaction network, without requiring parameters being negligible or the presence of an exact ACR motif. We find that aACR is far more common than classical ACR, as demonstrated in systems such as the \textit{Escherichia coli} EnvZ-OmpR system and MAPK signaling cascade. Furthermore, we mathematically prove that such ubiquity stems solely from network structures. Finally, we reveal a counterintuitive feature of aACR in systems with multiple conserved quantities, revealing subtle distinctions in how robustness manifests in complex biochemical networks. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:27:34 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hong",
"Hyukpyo",
""
],
[
"La Luz",
"Diego Rojas",
""
],
[
"Craciun",
"Gheorghe",
""
]
] |
2505.21901 | Zhixing Huang | Zhixing Huang, Bing Xue, Mengjie Zhang, Jeremy S. Ronney, Keith C. Gordon, and Daniel P. Killeen | Symbolically Regressing Fish Biomass Spectral Data: A Linear Genetic Programming Method with Tunable Primitives | null | null | null | null | cs.NE | http://creativecommons.org/licenses/by/4.0/ | Machine learning techniques play an important role in analyzing spectral data. The spectral data of fish biomass is useful in fish production, as it carries many important chemistry properties of fish meat. However, it is challenging for existing machine learning techniques to comprehensively discover hidden patterns from fish biomass spectral data since the spectral data often have a lot of noises while the training data are quite limited. To better analyze fish biomass spectral data, this paper models it as a symbolic regression problem and solves it by a linear genetic programming method with newly proposed tunable primitives. In the symbolic regression problem, linear genetic programming automatically synthesizes regression models based on the given primitives and training data. The tunable primitives further improve the approximation ability of the regression models by tuning their inherent coefficients. Our empirical results over ten fish biomass targets show that the proposed method improves the overall performance of fish biomass composition prediction. The synthesized regression models are compact and have good interpretability, which allow us to highlight useful features over the spectrum. Our further investigation also verifies the good generality of the proposed method across various spectral data treatments and other symbolic regression problems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:27:49 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Huang",
"Zhixing",
""
],
[
"Xue",
"Bing",
""
],
[
"Zhang",
"Mengjie",
""
],
[
"Ronney",
"Jeremy S.",
""
],
[
"Gordon",
"Keith C.",
""
],
[
"Killeen",
"Daniel P.",
""
]
] |
2505.21902 | Congcong Wang | Zhenyu Jiang, Xuemei Lu, Congcong Wang, Yingjie Huang, Xiaoshen Kang, Suyu Xiao, Xiyuan Zhang and Xin Shi | Rise Time and Charge Collection Efficiency of Graphene-Optimized 4H-SiC PIN Detector | null | null | null | null | physics.ins-det | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Silicon carbide detectors exhibit good detection performance and are being considered for detection applications. However, the presence of surface electrode of detector limits the application of low-penetration particle detectors, photodetectors and heavy-ion detection. A graphene-optimized 4H-SiC detector has been fabricated to expand the application of SiC detectors.Its electrical properties and the charge collection performance of {\alpha} particles are reported. The effective doping concentration of lightly doped 4H-SiC epitaxial layer is about 4.5\times10^{13}cm^{-3}, approaching the limit of the lowest doping level by the SiC epitaxial growth technique. The rise time of the graphene-optimized ring electrode detector is reduced by 24% at 200 V, compared to ring electrode detector. The charge collection efficiency (CCE) of graphene-optimized 4H-SiC PIN is 99.22%. When the irradiation dose is 2\times10^{11} n_{eq}/cm^2, the irradiation has no significant impact on the rise time and uniformity of the rise time for the graphene-optimized 4H-SiC detectors. This study proves that graphene has a certain radiation resistance. Graphene-optimized 4H-SiC detectors can not only reduce the signal rise time, but also improve uniformity of signal rise time and stability of charge collection. This research will expand the application of graphene-based 4H-SiC detectors in fields such as low energy ions, X-ray, UV light detection, particle physics, medical dosimetry and heavy-ion detection. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:31:58 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Jiang",
"Zhenyu",
""
],
[
"Lu",
"Xuemei",
""
],
[
"Wang",
"Congcong",
""
],
[
"Huang",
"Yingjie",
""
],
[
"Kang",
"Xiaoshen",
""
],
[
"Xiao",
"Suyu",
""
],
[
"Zhang",
"Xiyuan",
""
],
[
"Shi",
"Xin",
""
]
] |
2505.21903 | Ruihao Zheng | Ruihao Zheng, Zhenkun Wang, Yin Wu, and Maoguo Gong | Enhanced Ideal Objective Vector Estimation for Evolutionary Multi-Objective Optimization | 24 pages, 17 figures, 13 tables | null | null | null | cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The ideal objective vector, which comprises the optimal values of the $m$ objective functions in an $m$-objective optimization problem, is an important concept in evolutionary multi-objective optimization. Accurate estimation of this vector has consistently been a crucial task, as it is frequently used to guide the search process and normalize the objective space. Prevailing estimation methods all involve utilizing the best value concerning each objective function achieved by the individuals in the current or accumulated population. However, this paper reveals that the population-based estimation method can only work on simple problems but falls short on problems with substantial bias. The biases in multi-objective optimization problems can be divided into three categories, and an analysis is performed to illustrate how each category hinders the estimation of the ideal objective vector. Subsequently, a set of test instances is proposed to quantitatively evaluate the impact of various biases on the ideal objective vector estimation method. Beyond that, a plug-and-play component called enhanced ideal objective vector estimation (EIE) is introduced for multi-objective evolutionary algorithms (MOEAs). EIE features adaptive and fine-grained searches over $m$ subproblems defined by the extreme weighted sum method. EIE finally outputs $m$ solutions that can well approximate the ideal objective vector. In the experiments, EIE is integrated into three representative MOEAs. To demonstrate the wide applicability of EIE, algorithms are tested not only on the newly proposed test instances but also on existing ones. The results consistently show that EIE improves the ideal objective vector estimation and enhances the MOEA's performance. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:36:45 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zheng",
"Ruihao",
""
],
[
"Wang",
"Zhenkun",
""
],
[
"Wu",
"Yin",
""
],
[
"Gong",
"Maoguo",
""
]
] |
2505.21905 | Mo Zhou | Mo Zhou, Keren Ye, Viraj Shah, Kangfu Mei, Mauricio Delbracio, Peyman Milanfar, Vishal M. Patel, Hossein Talebi | Reference-Guided Identity Preserving Face Restoration | null | null | null | null | cs.CV cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Preserving face identity is a critical yet persistent challenge in diffusion-based image restoration. While reference faces offer a path forward, existing reference-based methods often fail to fully exploit their potential. This paper introduces a novel approach that maximizes reference face utility for improved face restoration and identity preservation. Our method makes three key contributions: 1) Composite Context, a comprehensive representation that fuses multi-level (high- and low-level) information from the reference face, offering richer guidance than prior singular representations. 2) Hard Example Identity Loss, a novel loss function that leverages the reference face to address the identity learning inefficiencies found in the existing identity loss. 3) A training-free method to adapt the model to multi-reference inputs during inference. The proposed method demonstrably restores high-quality faces and achieves state-of-the-art identity preserving restoration on benchmarks such as FFHQ-Ref and CelebA-Ref-Test, consistently outperforming previous work. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:46:34 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhou",
"Mo",
""
],
[
"Ye",
"Keren",
""
],
[
"Shah",
"Viraj",
""
],
[
"Mei",
"Kangfu",
""
],
[
"Delbracio",
"Mauricio",
""
],
[
"Milanfar",
"Peyman",
""
],
[
"Patel",
"Vishal M.",
""
],
[
"Talebi",
"Hossein",
""
]
] |
2505.21906 | Zhongyi Zhou | Zhongyi Zhou, Yichen Zhu, Junjie Wen, Chaomin Shen, Yi Xu | Vision-Language-Action Model with Open-World Embodied Reasoning from Pretrained Knowledge | Project page: https://chatvla-2.github.io/ | null | null | null | cs.RO cs.AI cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Vision-language-action (VLA) models have emerged as the next generation of models in robotics. However, despite leveraging powerful pre-trained Vision-Language Models (VLMs), existing end-to-end VLA systems often lose key capabilities during fine-tuning as the model adapts to specific robotic tasks. We argue that a generalizable VLA model should retain and expand upon the VLM's core competencies: 1) Open-world embodied reasoning - the VLA should inherit the knowledge from VLM, i.e., recognize anything that the VLM can recognize, capable of solving math problems, possessing visual-spatial intelligence, 2) Reasoning following - effectively translating the open-world reasoning into actionable steps for the robot. In this work, we introduce ChatVLA-2, a novel mixture-of-expert VLA model coupled with a specialized three-stage training pipeline designed to preserve the VLM's original strengths while enabling actionable reasoning. To validate our approach, we design a math-matching task wherein a robot interprets math problems written on a whiteboard and picks corresponding number cards from a table to solve equations. Remarkably, our method exhibits exceptional mathematical reasoning and OCR capabilities, despite these abilities not being explicitly trained within the VLA. Furthermore, we demonstrate that the VLA possesses strong spatial reasoning skills, enabling it to interpret novel directional instructions involving previously unseen objects. Overall, our method showcases reasoning and comprehension abilities that significantly surpass state-of-the-art imitation learning methods such as OpenVLA, DexVLA, and pi-zero. This work represents a substantial advancement toward developing truly generalizable robotic foundation models endowed with robust reasoning capacities. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:48:42 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhou",
"Zhongyi",
""
],
[
"Zhu",
"Yichen",
""
],
[
"Wen",
"Junjie",
""
],
[
"Shen",
"Chaomin",
""
],
[
"Xu",
"Yi",
""
]
] |
2505.21907 | Saleh Afzoon | Saleh Afzoon, Zahra Jahanandish, Phuong Thao Huynh, Amin Beheshti, Usman Naseem | Modeling and Optimizing User Preferences in AI Copilots: A Comprehensive Survey and Taxonomy | null | null | null | null | cs.AI cs.CL cs.HC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | AI copilots, context-aware, AI-powered systems designed to assist users in tasks such as software development and content creation, are becoming integral to modern workflows. As these systems grow in capability and adoption, personalization has emerged as a cornerstone for ensuring usability, trust, and productivity. Central to this personalization is preference optimization: the ability of AI copilots to detect, interpret, and align with individual user preferences. While personalization techniques are well-established in domains like recommender systems and dialogue agents, their adaptation to interactive, real-time systems like AI copilots remains fragmented and underexplored. This survey addresses this gap by synthesizing research on how user preferences are captured, modeled, and refined within the design of AI copilots. We introduce a unified definition of AI copilots and propose a phase-based taxonomy of preference optimization strategies, structured around pre-interaction, mid-interaction, and post-interaction stages. We analyze techniques for acquiring preference signals, modeling user intent, and integrating feedback loops, highlighting both established approaches and recent innovations. By bridging insights from AI personalization, human-AI collaboration, and large language model adaptation, this survey provides a structured foundation for designing adaptive, preference-aware AI copilots. It offers a holistic view of the available preference resources, how they can be leveraged, and which technical approaches are most suited to each stage of system design. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:52:39 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Afzoon",
"Saleh",
""
],
[
"Jahanandish",
"Zahra",
""
],
[
"Huynh",
"Phuong Thao",
""
],
[
"Beheshti",
"Amin",
""
],
[
"Naseem",
"Usman",
""
]
] |
2505.21908 | Hanyin Wang | Hanyin Wang, Zhenbang Wu, Gururaj Kolar, Hariprasad Korsapati, Brian Bartlett, Bryan Hull, Jimeng Sun | Reinforcement Learning for Out-of-Distribution Reasoning in LLMs: An Empirical Study on Diagnosis-Related Group Coding | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Diagnosis-Related Group (DRG) codes are essential for hospital reimbursement and operations but require labor-intensive assignment. Large Language Models (LLMs) struggle with DRG coding due to the out-of-distribution (OOD) nature of the task: pretraining corpora rarely contain private clinical or billing data. We introduce DRG-Sapphire, which uses large-scale reinforcement learning (RL) for automated DRG coding from clinical notes. Built on Qwen2.5-7B and trained with Group Relative Policy Optimization (GRPO) using rule-based rewards, DRG-Sapphire introduces a series of RL enhancements to address domain-specific challenges not seen in previous mathematical tasks. Our model achieves state-of-the-art accuracy on the MIMIC-IV benchmark and generates physician-validated reasoning for DRG assignments, significantly enhancing explainability. Our study further sheds light on broader challenges of applying RL to knowledge-intensive, OOD tasks. We observe that RL performance scales approximately linearly with the logarithm of the number of supervised fine-tuning (SFT) examples, suggesting that RL effectiveness is fundamentally constrained by the domain knowledge encoded in the base model. For OOD tasks like DRG coding, strong RL performance requires sufficient knowledge infusion prior to RL. Consequently, scaling SFT may be more effective and computationally efficient than scaling RL alone for such tasks. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:54:07 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wang",
"Hanyin",
""
],
[
"Wu",
"Zhenbang",
""
],
[
"Kolar",
"Gururaj",
""
],
[
"Korsapati",
"Hariprasad",
""
],
[
"Bartlett",
"Brian",
""
],
[
"Hull",
"Bryan",
""
],
[
"Sun",
"Jimeng",
""
]
] |
2505.21909 | Jiawei Fu | Jiawei Fu and Donald P. Green | Causal Inference for Experiments with Latent Outcomes: Key Results and Their Implications for Design and Analysis | null | null | null | null | econ.EM stat.AP stat.ME | http://creativecommons.org/licenses/by/4.0/ | How should we analyze randomized experiments in which the main outcome is measured in multiple ways and each measure contains some degree of error? Since Costner (1971) and Bagozzi (1977), methodological discussions of experiments with latent outcomes have reviewed the modeling assumptions that are invoked when the quantity of interest is the average treatment effect (ATE) of a randomized intervention on a latent outcome that is measured with error. Many authors have proposed methods to estimate this ATE when multiple measures of an outcome are available. Despite this extensive literature, social scientists rarely use these modeling approaches when analyzing experimental data, perhaps because the surge of interest in experiments coincides with increased skepticism about the modeling assumptions that these methods invoke. The present paper takes a fresh look at the use of latent variable models to analyze experiments. Like the skeptics, we seek to minimize reliance on ad hoc assumptions that are not rooted in the experimental design and measurement strategy. At the same time, we think that some of the misgivings that are frequently expressed about latent variable models can be addressed by modifying the research design in ways that make the underlying assumptions defensible or testable. We describe modeling approaches that enable researchers to identify and estimate key parameters of interest, suggest ways that experimental designs can be augmented so as to make the modeling requirements more credible, and discuss empirical tests of key modeling assumptions. Simulations and an empirical application illustrate the gains in terms of precision and robustness. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:54:45 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Fu",
"Jiawei",
""
],
[
"Green",
"Donald P.",
""
]
] |
2505.21910 | Xianbiao Qi | Xianbiao Qi, Yelin He, Jiaquan Ye, Chun-Guang Li, Bojia Zi, Xili Dai, Qin Zou, Rong Xiao | Taming Transformer Without Using Learning Rate Warmup | This paper is published as a conference paper at ICLR 2025 | null | null | null | cs.LG cs.CV | http://creativecommons.org/licenses/by/4.0/ | Scaling Transformer to a large scale without using some technical tricks such as learning rate warump and using an obviously lower learning rate is an extremely challenging task, and is increasingly gaining more attention. In this paper, we provide a theoretical analysis for the process of training Transformer and reveal the rationale behind the model crash phenomenon in the training process, termed \textit{spectral energy concentration} of ${\bW_q}^{\top} \bW_k$, which is the reason for a malignant entropy collapse, where ${\bW_q}$ and $\bW_k$ are the projection matrices for the query and the key in Transformer, respectively. To remedy this problem, motivated by \textit{Weyl's Inequality}, we present a novel optimization strategy, \ie, making the weight updating in successive steps smooth -- if the ratio $\frac{\sigma_{1}(\nabla \bW_t)}{\sigma_{1}(\bW_{t-1})}$ is larger than a threshold, we will automatically bound the learning rate to a weighted multiple of $\frac{\sigma_{1}(\bW_{t-1})}{\sigma_{1}(\nabla \bW_t)}$, where $\nabla \bW_t$ is the updating quantity in step $t$. Such an optimization strategy can prevent spectral energy concentration to only a few directions, and thus can avoid malignant entropy collapse which will trigger the model crash. We conduct extensive experiments using ViT, Swin-Transformer and GPT, showing that our optimization strategy can effectively and stably train these Transformers without using learning rate warmup. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:55:28 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Qi",
"Xianbiao",
""
],
[
"He",
"Yelin",
""
],
[
"Ye",
"Jiaquan",
""
],
[
"Li",
"Chun-Guang",
""
],
[
"Zi",
"Bojia",
""
],
[
"Dai",
"Xili",
""
],
[
"Zou",
"Qin",
""
],
[
"Xiao",
"Rong",
""
]
] |
2505.21911 | Yiheng Lin | Yiheng Lin, Shifang Zhao, Ting Liu, Xiaochao Qu, Luoqi Liu, Yao Zhao, Yunchao Wei | AlignGen: Boosting Personalized Image Generation with Cross-Modality Prior Alignment | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Personalized image generation aims to integrate user-provided concepts into text-to-image models, enabling the generation of customized content based on a given prompt. Recent zero-shot approaches, particularly those leveraging diffusion transformers, incorporate reference image information through multi-modal attention mechanism. This integration allows the generated output to be influenced by both the textual prior from the prompt and the visual prior from the reference image. However, we observe that when the prompt and reference image are misaligned, the generated results exhibit a stronger bias toward the textual prior, leading to a significant loss of reference content. To address this issue, we propose AlignGen, a Cross-Modality Prior Alignment mechanism that enhances personalized image generation by: 1) introducing a learnable token to bridge the gap between the textual and visual priors, 2) incorporating a robust training strategy to ensure proper prior alignment, and 3) employing a selective cross-modal attention mask within the multi-modal attention mechanism to further align the priors. Experimental results demonstrate that AlignGen outperforms existing zero-shot methods and even surpasses popular test-time optimization approaches. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:57:55 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lin",
"Yiheng",
""
],
[
"Zhao",
"Shifang",
""
],
[
"Liu",
"Ting",
""
],
[
"Qu",
"Xiaochao",
""
],
[
"Liu",
"Luoqi",
""
],
[
"Zhao",
"Yao",
""
],
[
"Wei",
"Yunchao",
""
]
] |
2505.21912 | Marvin Limpijankit | Marvin Limpijankit, John Kender | Detecting Cultural Differences in News Video Thumbnails via Computational Aesthetics | null | null | 10.36190/2024.61 | null | cs.CY cs.CV | http://creativecommons.org/licenses/by/4.0/ | We propose a two-step approach for detecting differences in the style of images across sources of differing cultural affinity, where images are first clustered into finer visual themes based on content before their aesthetic features are compared. We test this approach on 2,400 YouTube video thumbnails taken equally from two U.S. and two Chinese YouTube channels, and relating equally to COVID-19 and the Ukraine conflict. Our results suggest that while Chinese thumbnails are less formal and more candid, U.S. channels tend to use more deliberate, proper photographs as thumbnails. In particular, U.S. thumbnails are less colorful, more saturated, darker, more finely detailed, less symmetric, sparser, less varied, and more up close and personal than Chinese thumbnails. We suggest that most of these differences reflect cultural preferences, and that our methods and observations can serve as a baseline against which suspected visual propaganda can be computed and compared. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:58:41 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Limpijankit",
"Marvin",
""
],
[
"Kender",
"John",
""
]
] |
2505.21913 | Mengqi Zhang | Mengqi Zhang | On the intracyclic instability in Stokes layers | 19 pages, 10 pages, accepted in Journal of Fluid Mechanics | null | null | null | physics.flu-dyn | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Time-dependent fluid dynamics plays a crucial role in both natural phenomena and industrial applications. Understanding the flow instabilities and transitions within these dynamical systems is essential for predicting and controlling their unsteady behaviour. A classic example of time-dependent flow is the Stokes layer. To study the transition mechanism in this flow, we employ the Finite-Time Lyapunov Exponent (FTLE) to demonstrate that a linear energy amplification mechanism may explain the intracyclic instability in the transitional Stokes layer, supported by favourable comparisons with experimental measurements of axial turbulence intensity. This complements existing theories applied to the Stokes layer in the literature, including the Floquet analysis and the instantaneous/momentary analyses, which have struggled to capture this experimental observation accurately. The FTLE analysis is closely related to the transient growth analysis, formulated as an optimisation problem of the disturbance energy growth over time. We found that the energy amplification weakens as the finite Stokes layer becomes more confined and the oscillating frequency has a non-monotonic effect on the maximum transient growth. Based on these results, we recommend future experimental studies to validate this linear mechanism. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:59:00 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhang",
"Mengqi",
""
]
] |
2505.21914 | Qi Wu | Chenfeng Wei, Qi Wu, Si Zuo, Jiahua Xu, Boyang Zhao, Zeyu Yang, Guotao Xie, Shenhong Wang | LiDARDustX: A LiDAR Dataset for Dusty Unstructured Road Environments | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Autonomous driving datasets are essential for validating the progress of intelligent vehicle algorithms, which include localization, perception, and prediction. However, existing datasets are predominantly focused on structured urban environments, which limits the exploration of unstructured and specialized scenarios, particularly those characterized by significant dust levels. This paper introduces the LiDARDustX dataset, which is specifically designed for perception tasks under high-dust conditions, such as those encountered in mining areas. The LiDARDustX dataset consists of 30,000 LiDAR frames captured by six different LiDAR sensors, each accompanied by 3D bounding box annotations and point cloud semantic segmentation. Notably, over 80% of the dataset comprises dust-affected scenes. By utilizing this dataset, we have established a benchmark for evaluating the performance of state-of-the-art 3D detection and segmentation algorithms. Additionally, we have analyzed the impact of dust on perception accuracy and delved into the causes of these effects. The data and further information can be accessed at: https://github.com/vincentweikey/LiDARDustX. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 02:59:19 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wei",
"Chenfeng",
""
],
[
"Wu",
"Qi",
""
],
[
"Zuo",
"Si",
""
],
[
"Xu",
"Jiahua",
""
],
[
"Zhao",
"Boyang",
""
],
[
"Yang",
"Zeyu",
""
],
[
"Xie",
"Guotao",
""
],
[
"Wang",
"Shenhong",
""
]
] |
2505.21915 | Mir Sazzat Hossain | Mir Sazzat Hossain, Ovi Paul, Md Akil Raihan Iftee, Rakibul Hasan Rajib, Abu Bakar Siddik Nayem, Anis Sarker, Arshad Momen, Md. Ashraful Amin, Amin Ahsan Ali, AKM Mahbubur Rahman | BD Open LULC Map: High-resolution land use land cover mapping & benchmarking for urban development in Dhaka, Bangladesh | 6 pages, 5 figures, 3 tables, Accepted In ICIP 2025 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Land Use Land Cover (LULC) mapping using deep learning significantly enhances the reliability of LULC classification, aiding in understanding geography, socioeconomic conditions, poverty levels, and urban sprawl. However, the scarcity of annotated satellite data, especially in South/East Asian developing countries, poses a major challenge due to limited funding, diverse infrastructures, and dense populations. In this work, we introduce the BD Open LULC Map (BOLM), providing pixel-wise LULC annotations across eleven classes (e.g., Farmland, Water, Forest, Urban Structure, Rural Built-Up) for Dhaka metropolitan city and its surroundings using high-resolution Bing satellite imagery (2.22 m/pixel). BOLM spans 4,392 sq km (891 million pixels), with ground truth validated through a three-stage process involving GIS experts. We benchmark LULC segmentation using DeepLab V3+ across five major classes and compare performance on Bing and Sentinel-2A imagery. BOLM aims to support reliable deep models and domain adaptation tasks, addressing critical LULC dataset gaps in South/East Asia. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:00:03 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hossain",
"Mir Sazzat",
""
],
[
"Paul",
"Ovi",
""
],
[
"Iftee",
"Md Akil Raihan",
""
],
[
"Rajib",
"Rakibul Hasan",
""
],
[
"Nayem",
"Abu Bakar Siddik",
""
],
[
"Sarker",
"Anis",
""
],
[
"Momen",
"Arshad",
""
],
[
"Amin",
"Md. Ashraful",
""
],
[
"Ali",
"Amin Ahsan",
""
],
[
"Rahman",
"AKM Mahbubur",
""
]
] |
2505.21916 | Yihang Hu | Yihang Hu, Pingyue Sheng, Shengjie Wang, Yang Gao | Mastering Agile Tasks with Limited Trials | null | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Embodied robots nowadays can already handle many real-world manipulation tasks. However, certain other real-world tasks (e.g., shooting a basketball into a hoop) are highly agile and require high execution precision, presenting additional challenges for methods primarily designed for quasi-static manipulation tasks. This leads to increased efforts in costly data collection, laborious reward design, or complex motion planning. Such tasks, however, are far less challenging for humans. Say a novice basketball player typically needs only $\sim$10 attempts to make their first successful shot, by roughly imitating a motion prior and then iteratively adjusting their motion based on the past outcomes. Inspired by this human learning paradigm, we propose the Adaptive Diffusion Action Plannin (ADAP) algorithm, a simple & scalable approach which iteratively refines its action plan by few real-world trials within a learned prior motion pattern, until reaching a specific goal. Experiments demonstrated that ADAP can learn and accomplish a wide range of goal-conditioned agile dynamic tasks with human-level precision and efficiency directly in real-world, such as throwing a basketball into the hoop in fewer than 10 trials. Project website:https://adap-robotics.github.io/ . | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:03:38 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hu",
"Yihang",
""
],
[
"Sheng",
"Pingyue",
""
],
[
"Wang",
"Shengjie",
""
],
[
"Gao",
"Yang",
""
]
] |
2505.21917 | Ryan Schneider | James Demmel, Ioana Dumitriu, and Ryan Schneider | Structured Divide-and-Conquer for the Definite Generalized Eigenvalue Problem | 27 pages, 3 figures | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper presents a fast, randomized divide-and-conquer algorithm for the definite generalized eigenvalue problem, which corresponds to pencils $(A,B)$ in which $A$ and $B$ are Hermitian and the Crawford number $\gamma(A,B) = \min_{||x||_2 = 1} |x^H(A+iB)x|$ is positive. Adapted from the fastest known method for diagonalizing arbitrary matrix pencils [Foundations of Computational Mathematics 2024], the algorithm is both inverse-free and highly parallel. As in the general case, randomization takes the form of perturbations applied to the input matrices, which regularize the problem for compatibility with fast, divide-and-conquer eigensolvers -- i.e., the now well-established phenomenon of pseudospectral shattering. We demonstrate that this high-level approach to diagonalization can be executed in a structure-aware fashion by (1) extending pseudospectral shattering to definite pencils under structured perturbations (either random diagonal or sampled from the Gaussian Unitary Ensemble) and (2) formulating the divide-and-conquer procedure in a way that maintains definiteness. The result is a specialized solver whose complexity, when applied to definite pencils, is provably lower than that of general divide-and-conquer. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:04:06 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Demmel",
"James",
""
],
[
"Dumitriu",
"Ioana",
""
],
[
"Schneider",
"Ryan",
""
]
] |
2505.21918 | Tsuyoshi Okita | Haruki Kai and Tsuyoshi Okita | Self-supervised Learning Method Using Transformer for Multi-dimensional Sensor Data Processing | 25 pages, 4 figures | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We developed a deep learning algorithm for human activity recognition using sensor signals as input. In this study, we built a pretrained language model based on the Transformer architecture, which is widely used in natural language processing. By leveraging this pretrained model, we aimed to improve performance on the downstream task of human activity recognition. While this task can be addressed using a vanilla Transformer, we propose an enhanced n-dimensional numerical processing Transformer that incorporates three key features: embedding n-dimensional numerical data through a linear layer, binning-based pre-processing, and a linear transformation in the output layer. We evaluated the effectiveness of our proposed model across five different datasets. Compared to the vanilla Transformer, our model demonstrated 10%-15% improvements in accuracy. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:04:13 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Kai",
"Haruki",
""
],
[
"Okita",
"Tsuyoshi",
""
]
] |
2505.21919 | Yue Zhu | Yue Zhu, Hao Yu, Chen Wang, Zhuoran Liu, Eun Kyung Lee | Towards Efficient Key-Value Cache Management for Prefix Prefilling in LLM Inference | This paper has been accepted at IEEE Cloud 2025 as WIP paper. The final version will appear in IEEE Xplore | null | null | null | cs.ET cs.AI cs.DC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The increasing adoption of large language models (LLMs) with extended context windows necessitates efficient Key-Value Cache (KVC) management to optimize inference performance. Inference workloads like Retrieval-Augmented Generation (RAG) and agents exhibit high cache reusability, making efficient caching critical to reducing redundancy and improving speed. We analyze real-world KVC access patterns using publicly available traces and evaluate commercial key-value stores like Redis and state-of-the-art RDMA-based systems (CHIME [1] and Sherman [2]) for KVC metadata management. Our work demonstrates the lack of tailored storage solution for KVC prefilling, underscores the need for an efficient distributed caching system with optimized metadata management for LLM workloads, and provides insights into designing improved KVC management systems for scalable, low-latency inference. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:05:55 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhu",
"Yue",
""
],
[
"Yu",
"Hao",
""
],
[
"Wang",
"Chen",
""
],
[
"Liu",
"Zhuoran",
""
],
[
"Lee",
"Eun Kyung",
""
]
] |
2505.21920 | Yuanhong Zhang | Yuanhong Zhang, Muyao Yuan, Weizhan Zhang, Tieliang Gong, Wen Wen, Jiangyong Ying, Weijie Shi | InfoSAM: Fine-Tuning the Segment Anything Model from An Information-Theoretic Perspective | Accepted by ICML 2025 (Highlight) | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Segment Anything Model (SAM), a vision foundation model, exhibits impressive zero-shot capabilities in general tasks but struggles in specialized domains. Parameter-efficient fine-tuning (PEFT) is a promising approach to unleash the potential of SAM in novel scenarios. However, existing PEFT methods for SAM neglect the domain-invariant relations encoded in the pre-trained model. To bridge this gap, we propose InfoSAM, an information-theoretic approach that enhances SAM fine-tuning by distilling and preserving its pre-trained segmentation knowledge. Specifically, we formulate the knowledge transfer process as two novel mutual information-based objectives: (i) to compress the domain-invariant relation extracted from pre-trained SAM, excluding pseudo-invariant information as possible, and (ii) to maximize mutual information between the relational knowledge learned by the teacher (pre-trained SAM) and the student (fine-tuned model). The proposed InfoSAM establishes a robust distillation framework for PEFT of SAM. Extensive experiments across diverse benchmarks validate InfoSAM's effectiveness in improving SAM family's performance on real-world tasks, demonstrating its adaptability and superiority in handling specialized scenarios. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:09:22 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhang",
"Yuanhong",
""
],
[
"Yuan",
"Muyao",
""
],
[
"Zhang",
"Weizhan",
""
],
[
"Gong",
"Tieliang",
""
],
[
"Wen",
"Wen",
""
],
[
"Ying",
"Jiangyong",
""
],
[
"Shi",
"Weijie",
""
]
] |
2505.21921 | Yuan Miao | Yuan Miao, Linhao Li, Hosho Katsura, Masahito Yamazaki | Exact Quantum Many-Body Scars in 2D Quantum Gauge Models | 35 pages, 17 figures | null | null | null | cond-mat.str-el cond-mat.stat-mech quant-ph | http://creativecommons.org/licenses/by/4.0/ | Quantum many-body scars (QMBS) serve as important examples of ergodicity-breaking phenomena in quantum many-body systems. Despite recent extensive studies, exact QMBS are rare in dimensions higher than one. In this paper, we study a two-dimensional quantum $\mathbb{Z}_2$ gauge model that is dual to a two-dimensional spin-$1/2$ XY model defined on bipartite graphs. We identify the exact eigenstates of the XY model with a tower structure as exact QMBS. Exploiting the duality transformation, we show that the exact QMBS of the XY model (and XXZ model) after the transformation are the exact QMBS of the dual $\mathbb{Z}_2$ gauge model. This construction is versatile and has potential applications for finding new QMBS in other higher-dimensional models. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:10:22 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Miao",
"Yuan",
""
],
[
"Li",
"Linhao",
""
],
[
"Katsura",
"Hosho",
""
],
[
"Yamazaki",
"Masahito",
""
]
] |
2505.21922 | Juseung Oh | Juseung Oh, Wontaek Kim, Gyouil Jeong, Yeri Lee, Jihun Kim, Hyeongjoon Kim, Hyeon Suk Shin and Sunmin Ryu | Dual-Polarization SHG Interferometry for Imaging Antiparallel Domains and Stacking Angles of 2D Heterocrystals | 26 pages, 5 figures | null | null | null | physics.optics cond-mat.mes-hall cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Optical second-harmonic generation (SHG) enables orientational polarimetry for crystallographic analysis and domain imaging of various materials. However, conventional intensity polarimetry, which neglects phase information, fails to resolve antiparallel domains and to describe two-dimensional heterostructures, which represent a new class of van der Waals-bound composite crystals. In this work, we report dual-polarization spectral phase interferometry (DP-SPI) and establish a generalized SHG superposition model that incorporates the observables of DP-SPI. Antiparallel domains of monolayer transition metal dichalcogenides (TMDs) were successfully imaged with distinction, validating the interferometric polarimetry. From DP interferograms of TMD heterobilayers, the orientation of each layer could be determined, enabling layer-resolved probing. By employing the superposition model, we also demonstrate the photonic design and fabrication of ternary TMD heterostructures for circularly polarized SHG. These methods, providing comprehensive SHG measurements and theoretical description, can be extended to heterostructures consisting of more than two constituent layers and are not limited to TMDs or 2D materials. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:12:34 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Oh",
"Juseung",
""
],
[
"Kim",
"Wontaek",
""
],
[
"Jeong",
"Gyouil",
""
],
[
"Lee",
"Yeri",
""
],
[
"Kim",
"Jihun",
""
],
[
"Kim",
"Hyeongjoon",
""
],
[
"Shin",
"Hyeon Suk",
""
],
[
"Ryu",
"Sunmin",
""
]
] |
2505.21923 | Asal Mehradfar | Asal Mehradfar, Xuzhe Zhao, Yilun Huang, Emir Ceyani, Yankai Yang, Shihao Han, Hamidreza Aghasi, Salman Avestimehr | FALCON: An ML Framework for Fully Automated Layout-Constrained Analog Circuit Design | null | null | null | null | cs.LG cs.AI cs.AR cs.CE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Designing analog circuits from performance specifications is a complex, multi-stage process encompassing topology selection, parameter inference, and layout feasibility. We introduce FALCON, a unified machine learning framework that enables fully automated, specification-driven analog circuit synthesis through topology selection and layout-constrained optimization. Given a target performance, FALCON first selects an appropriate circuit topology using a performance-driven classifier guided by human design heuristics. Next, it employs a custom, edge-centric graph neural network trained to map circuit topology and parameters to performance, enabling gradient-based parameter inference through the learned forward model. This inference is guided by a differentiable layout cost, derived from analytical equations capturing parasitic and frequency-dependent effects, and constrained by design rules. We train and evaluate FALCON on a large-scale custom dataset of 1M analog mm-wave circuits, generated and simulated using Cadence Spectre across 20 expert-designed topologies. Through this evaluation, FALCON demonstrates >99\% accuracy in topology inference, <10\% relative error in performance prediction, and efficient layout-aware design that completes in under 1 second per instance. Together, these results position FALCON as a practical and extensible foundation model for end-to-end analog circuit design automation. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:16:08 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Mehradfar",
"Asal",
""
],
[
"Zhao",
"Xuzhe",
""
],
[
"Huang",
"Yilun",
""
],
[
"Ceyani",
"Emir",
""
],
[
"Yang",
"Yankai",
""
],
[
"Han",
"Shihao",
""
],
[
"Aghasi",
"Hamidreza",
""
],
[
"Avestimehr",
"Salman",
""
]
] |
2505.21924 | Matthew Sfeir | Kamyar Rashidi and Matthew Y. Sfeir | Modified Transfer Matrix Method for the Extraction of Material Properties via Terahertz Time-Domain Spectroscopy | null | null | null | null | physics.optics cond-mat.mes-hall cond-mat.mtrl-sci physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | Terahertz Time-Domain Spectroscopy is a powerful technique for extracting the low-frequency optical properties of materials. However, the optical constants are difficult to determine directly from the experimental transfer function, such that various numerical approximations must be implemented to describe specific conditions. Here, we introduce a modified Transfer Matrix Method that uses a genetic algorithm for optimization to determine the refractive index of materials in the THz regime. We show that this approach is generally applicable across a wide range of refractive indices, structures, and frequency ranges. Our method is intuitive and yields accurate results compared to leading methods across a wide spectral range (1-15 THz). | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:19:35 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Rashidi",
"Kamyar",
""
],
[
"Sfeir",
"Matthew Y.",
""
]
] |
2505.21925 | Chong Zeng | Chong Zeng and Yue Dong and Pieter Peers and Hongzhi Wu and Xin Tong | RenderFormer: Transformer-based Neural Rendering of Triangle Meshes with Global Illumination | Accepted to SIGGRAPH 2025. Project page: https://microsoft.github.io/renderformer | ACM SIGGRAPH 2025 Conference Papers | 10.1145/3721238.3730595 | null | cs.GR cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present RenderFormer, a neural rendering pipeline that directly renders an image from a triangle-based representation of a scene with full global illumination effects and that does not require per-scene training or fine-tuning. Instead of taking a physics-centric approach to rendering, we formulate rendering as a sequence-to-sequence transformation where a sequence of tokens representing triangles with reflectance properties is converted to a sequence of output tokens representing small patches of pixels. RenderFormer follows a two stage pipeline: a view-independent stage that models triangle-to-triangle light transport, and a view-dependent stage that transforms a token representing a bundle of rays to the corresponding pixel values guided by the triangle-sequence from the view-independent stage. Both stages are based on the transformer architecture and are learned with minimal prior constraints. We demonstrate and evaluate RenderFormer on scenes with varying complexity in shape and light transport. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:20:46 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zeng",
"Chong",
""
],
[
"Dong",
"Yue",
""
],
[
"Peers",
"Pieter",
""
],
[
"Wu",
"Hongzhi",
""
],
[
"Tong",
"Xin",
""
]
] |
2505.21926 | Yin Hua | Yin Hua, Zhiqiang Liu, Mingyang Chen, Zheng Fang, Chi Man Wong, Lingxiao Li, Chi Man Vong, Huajun Chen, Wen Zhang | Beyond Completion: A Foundation Model for General Knowledge Graph Reasoning | ACL 2025 Findings | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | In natural language processing (NLP) and computer vision (CV), the successful application of foundation models across diverse tasks has demonstrated their remarkable potential. However, despite the rich structural and textual information embedded in knowledge graphs (KGs), existing research of foundation model for KG has primarily focused on their structural aspects, with most efforts restricted to in-KG tasks (e.g., knowledge graph completion, KGC). This limitation has hindered progress in addressing more challenging out-of-KG tasks. In this paper, we introduce MERRY, a foundation model for general knowledge graph reasoning, and investigate its performance across two task categories: in-KG reasoning tasks (e.g., KGC) and out-of-KG tasks (e.g., KG question answering, KGQA). We not only utilize the structural information, but also the textual information in KGs. Specifically, we propose a multi-perspective Conditional Message Passing (CMP) encoding architecture to bridge the gap between textual and structural modalities, enabling their seamless integration. Additionally, we introduce a dynamic residual fusion module to selectively retain relevant textual information and a flexible edge scoring mechanism to adapt to diverse downstream tasks. Comprehensive evaluations on 28 datasets demonstrate that MERRY outperforms existing baselines in most scenarios, showcasing strong reasoning capabilities within KGs and excellent generalization to out-of-KG tasks such as KGQA. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:21:28 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hua",
"Yin",
""
],
[
"Liu",
"Zhiqiang",
""
],
[
"Chen",
"Mingyang",
""
],
[
"Fang",
"Zheng",
""
],
[
"Wong",
"Chi Man",
""
],
[
"Li",
"Lingxiao",
""
],
[
"Vong",
"Chi Man",
""
],
[
"Chen",
"Huajun",
""
],
[
"Zhang",
"Wen",
""
]
] |
2505.21927 | Yu Cao | Hong-Lin Lin, Pragati Aashna, Yu Cao, Aaron Danner | Giant electro-optic coefficient in single crystal barium titanate on oxide insulator based Mach-Zehnder interferometer | null | null | null | null | physics.optics physics.app-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Electro-optic modulators are indispensable components of modern day photonic integrated circuits (PICs). Recently lithium niobate has emerged as a key material to realize large-bandwidth high-speed modulation, but next-generation modulators require high-density integration, low cost, low power and high performance simultaneously, which are difficult to achieve with established integrated lithium niobate photonics platforms due to its limited electro-optic coefficient. Leveraging its exceptional Pockels coefficient, barium titanate (BTO) in the thin film form has emerged as a promising alternative but the electro-optic coefficients reported in thin-film BTO often fall short of bulk values due to challenges in film growth and waveguide fabrication. Here, we report, to the best of our knowledge, the largest Pockels coefficient (r42) of 1268 pm/V in thin film BTO platform, which is very close to the bulk value. We measure it by using an unbalanced Mach-Zehnder interferometer, fabricated by an optimized wet-etching method for realising single-mode waveguides in single-crystal barium titanate-on-insulator grown by pulsed laser deposition. This giant r42 is extracted from a device in which the optical mode is fully confined within a single-crystal BTO waveguide. This approach contrasts with previous designs where the core material - typically silicon or silicon nitride - supports only partial confinement, resulting in an evanescent overlap with a multi-crystalline BTO layer. This highly confined BTO on insulator electro-optic modulation technology may significantly advance the field of ultra-low-power integrated photonic devices and allows for the realization of next-generation efficient and compact photonic circuits. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:22:01 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lin",
"Hong-Lin",
""
],
[
"Aashna",
"Pragati",
""
],
[
"Cao",
"Yu",
""
],
[
"Danner",
"Aaron",
""
]
] |
2505.21928 | Lianghui Zhu | Lianghui Zhu, Xitong Ling, Minxi Ouyang, Xiaoping Liu, Mingxi Fu, Tian Guan, Fanglei Fu, Xuanyu Wang, Maomao Zeng, Mingxi Zhu, Yibo Jin, Liming Liu, Song Duan, Qiming He, Yizhi Wang, Luxi Xie, Houqiang Li, Yonghong He, Sufang Tian | Subspecialty-Specific Foundation Model for Intelligent Gastrointestinal Pathology | null | null | null | null | eess.IV cs.AI cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Gastrointestinal (GI) diseases represent a clinically significant burden, necessitating precise diagnostic approaches to optimize patient outcomes. Conventional histopathological diagnosis, heavily reliant on the subjective interpretation of pathologists, suffers from limited reproducibility and diagnostic variability. To overcome these limitations and address the lack of pathology-specific foundation models for GI diseases, we develop Digepath, a specialized foundation model for GI pathology. Our framework introduces a dual-phase iterative optimization strategy combining pretraining with fine-screening, specifically designed to address the detection of sparsely distributed lesion areas in whole-slide images. Digepath is pretrained on more than 353 million image patches from over 200,000 hematoxylin and eosin-stained slides of GI diseases. It attains state-of-the-art performance on 33 out of 34 tasks related to GI pathology, including pathological diagnosis, molecular prediction, gene mutation prediction, and prognosis evaluation, particularly in diagnostically ambiguous cases and resolution-agnostic tissue classification.We further translate the intelligent screening module for early GI cancer and achieve near-perfect 99.6% sensitivity across 9 independent medical institutions nationwide. The outstanding performance of Digepath highlights its potential to bridge critical gaps in histopathological practice. This work not only advances AI-driven precision pathology for GI diseases but also establishes a transferable paradigm for other pathology subspecialties. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:22:08 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhu",
"Lianghui",
""
],
[
"Ling",
"Xitong",
""
],
[
"Ouyang",
"Minxi",
""
],
[
"Liu",
"Xiaoping",
""
],
[
"Fu",
"Mingxi",
""
],
[
"Guan",
"Tian",
""
],
[
"Fu",
"Fanglei",
""
],
[
"Wang",
"Xuanyu",
""
],
[
"Zeng",
"Maomao",
""
],
[
"Zhu",
"Mingxi",
""
],
[
"Jin",
"Yibo",
""
],
[
"Liu",
"Liming",
""
],
[
"Duan",
"Song",
""
],
[
"He",
"Qiming",
""
],
[
"Wang",
"Yizhi",
""
],
[
"Xie",
"Luxi",
""
],
[
"Li",
"Houqiang",
""
],
[
"He",
"Yonghong",
""
],
[
"Tian",
"Sufang",
""
]
] |
2505.21929 | Linhui Wu | Linhui Wu, Fu-Guo Xie, Qian Zheng, Quan Guo, Huanyuan Shan, Dan Hu, Stefan W. Duchesne, Nicholas Seymour, Jingying Wang, Junhua Gu, Qingwen Wu, Zhenghao Zhu, Melanie Johnston-Hollitt, Chris Riseley, Xu-Liang Fan | MWA and VLA Observations of Diffuse Radio Lobes in M 87 | 22 pages, 11 figures; accepted by ApJ | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by-nc-sa/4.0/ | This study investigates the projected, quasi-symmetric $\sim\rm46\,kpc$-scale diffuse radio lobes surrounding the giant elliptical galaxy M\,87, utilizing well-sampled wideband ($\rm 60\,MHz-10.55\,GHz$) observations from MWA and VLA, supplemented by data from LOFAR and Effelsberg. The observed structures feature sharp edges and filaments, with nearly uniform and moderately steep spectral indices ($\alpha$, mostly within $-1.2\leq\alpha\leq-0.8$), indicating turbulence. Well-sampled radio spectra for the lobes' diffuse region are derived using the continuous injection (CI) model (with $\alpha_{\rm inj}\simeq-0.86$ and $\nu_{\rm b}\simeq1.72\rm\,GHz$), and for its three localized regions using the impulsive injection model (e.g., JP model). From energy equipartition analysis, we estimate the typical magnetic field strength in the lobes' diffuse region to be $B_{\rm eq}\simeq10\,\mu\rm G$. The age of the lobes is estimated as $\sim30-50\,\rm~Myr$, based on lifetimes derived from the CI and JP models and sound crossing time. Outflow powers of $\sim(0.2-2)\times10^{44}\,\rm erg\,s^{-1}$ for the lobes' diffuse components and $\sim(1-11)\times10^{44}\,\rm erg\,s^{-1}$ for the whole source are calculated. With this power assessment, we conclude that the galactic stellar wind has a negligible effect, the active galactic nucleus (AGN)-driven jet can provide the necessary energy for the whole system. Furthermore, we argue that while the wind driven by current AGN activity is unlikely to power the lobes' diffuse components, an average enhancement of AGN activity by a factor of $\sim 10^2$ over the past $\sim 30-50$ Myr remains plausible. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:23:31 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wu",
"Linhui",
""
],
[
"Xie",
"Fu-Guo",
""
],
[
"Zheng",
"Qian",
""
],
[
"Guo",
"Quan",
""
],
[
"Shan",
"Huanyuan",
""
],
[
"Hu",
"Dan",
""
],
[
"Duchesne",
"Stefan W.",
""
],
[
"Seymour",
"Nicholas",
""
],
[
"Wang",
"Jingying",
""
],
[
"Gu",
"Junhua",
""
],
[
"Wu",
"Qingwen",
""
],
[
"Zhu",
"Zhenghao",
""
],
[
"Johnston-Hollitt",
"Melanie",
""
],
[
"Riseley",
"Chris",
""
],
[
"Fan",
"Xu-Liang",
""
]
] |
2505.21930 | Ziniu Zhang | Dongyue Li, Ziniu Zhang, Lu Wang, and Hongyang R. Zhang | Efficient Ensemble for Fine-tuning Language Models on Multiple Datasets | 17 pages. To appear in ACL'25 | null | null | null | cs.LG cs.CL | http://creativecommons.org/licenses/by/4.0/ | This paper develops an ensemble method for fine-tuning a language model to multiple datasets. Existing methods, such as quantized LoRA (QLoRA), are efficient when adapting to a single dataset. When training on multiple datasets of different tasks, a common setup in practice, it remains unclear how to design an efficient adaptation for fine-tuning language models. We propose to use an ensemble of multiple smaller adapters instead of a single adapter per task. We design an efficient algorithm that partitions $n$ datasets into $m$ groups, where $m$ is typically much smaller than $n$ in practice, and train one adapter for each group before taking a weighted combination to form the ensemble. The algorithm leverages a first-order approximation property of low-rank adaptation to quickly obtain the fine-tuning performances of dataset combinations since methods like LoRA stay close to the base model. Hence, we use the gradients of the base model to estimate its behavior during fine-tuning. Empirically, this approximation holds with less than $1\%$ error on models with up to $34$ billion parameters, leading to an estimation of true fine-tuning performances under $5\%$ error while speeding up computation compared to base fine-tuning by $105$ times. When applied to fine-tune Llama and GPT models on ten text classification tasks, our approach provides up to $10\%$ higher average test accuracy over QLoRA, with only $9\%$ more FLOPs. On a Llama model with $34$ billion parameters, an ensemble of QLoRA increases test accuracy by $3\%$ compared to QLoRA, with only $8\%$ more FLOPs. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:27:08 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Li",
"Dongyue",
""
],
[
"Zhang",
"Ziniu",
""
],
[
"Wang",
"Lu",
""
],
[
"Zhang",
"Hongyang R.",
""
]
] |
2505.21931 | Sina Mohammadi | Sina Mohammadi, Ali Hassan, Rouzbeh Haghighi, Van-Hai Bui and Wencong Su | Large Language Models for Solving Economic Dispatch Problem | 5 pages, 3 figures, Accepted, 2025 IEEE Energy Conversion Conference and Expo (ECCE 2025), Philadelphia, PA | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | This paper investigates the capability of off-the-shelf large language models (LLMs) to solve the economic dispatch (ED) problem. ED is a hard-constrained optimization problem solved on a day-ahead timescale by grid operators to minimize electricity generation costs while accounting for physical and engineering constraints. Numerous approaches have been proposed, but these typically require either mathematical formulations, face convergence issues, or depend on extensive labeled data and training time. This work implements LLMs enhanced with reasoning capabilities to address the classic lossless ED problem. The proposed approach avoids the need for explicit mathematical formulations, does not suffer from convergence challenges, and requires neither labeled data nor extensive training. A few-shot learning technique is utilized in two different prompting contexts. The IEEE 118-bus system with 19 generation units serves as the evaluation benchmark. Results demonstrate that various prompting strategies enable LLMs to effectively solve the ED problem, offering a convenient and efficient alternative. Consequently, this approach presents a promising future solution for ED tasks, particularly when foundational power system models are available. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:31:54 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Mohammadi",
"Sina",
""
],
[
"Hassan",
"Ali",
""
],
[
"Haghighi",
"Rouzbeh",
""
],
[
"Bui",
"Van-Hai",
""
],
[
"Su",
"Wencong",
""
]
] |
2505.21932 | Adriana Duncan | Adriana L. Duncan, Joe Kileel | Higher-Order Group Synchronization | 40 pages | null | null | null | stat.ML cs.CV cs.LG math.CO math.OC | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Group synchronization is the problem of determining reliable global estimates from noisy local measurements on networks. The typical task for group synchronization is to assign elements of a group to the nodes of a graph in a way that respects group elements given on the edges which encode information about local pairwise relationships between the nodes. In this paper, we introduce a novel higher-order group synchronization problem which operates on a hypergraph and seeks to synchronize higher-order local measurements on the hyperedges to obtain global estimates on the nodes. Higher-order group synchronization is motivated by applications to computer vision and image processing, among other computational problems. First, we define the problem of higher-order group synchronization and discuss its mathematical foundations. Specifically, we give necessary and sufficient synchronizability conditions which establish the importance of cycle consistency in higher-order group synchronization. Then, we propose the first computational framework for general higher-order group synchronization; it acts globally and directly on higher-order measurements using a message passing algorithm. We discuss theoretical guarantees for our framework, including convergence analyses under outliers and noise. Finally, we show potential advantages of our method through numerical experiments. In particular, we show that in certain cases our higher-order method applied to rotational and angular synchronization outperforms standard pairwise synchronization methods and is more robust to outliers. We also show that our method has comparable performance on simulated cryo-electron microscopy (cryo-EM) data compared to a standard cryo-EM reconstruction package. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:37:10 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Duncan",
"Adriana L.",
""
],
[
"Kileel",
"Joe",
""
]
] |
2505.21933 | Raymon Watson | R. S. Watson and K. V. Kheruntsyan | Universal principles for sudden-quench quantum Otto engines | 12 pages, 6 figures | null | null | null | cond-mat.quant-gas cond-mat.stat-mech quant-ph | http://creativecommons.org/licenses/by/4.0/ | We apply a simple sudden quench approximation for the unitary work strokes of a quantum Otto engine in order to provide a general analysis of its performance, applicable to arbitrary quantum models with two-body interactions. This work extends recent results for an interaction-driven Otto cycle to generic many-body interacting quantum models, providing universal bounds on their operation efficiency. From this, we demonstrate that the net work of such an engine cycle is determined entirely by interparticle correlations. Applications are demonstrated for a handful of paradigmatic many-body quantum models, including a novel engine -- with a spin-1/2 Fermi gas with contact two-body interactions as its working medium -- in which we leverage control over spin polarization to greatly enhance its performance compared to the unpolarized case. We then extend the analysis of interaction-driven quantum Otto engine cycles to systems where control is exerted over the strength of arbitrary quantum operators that might be present in the system Hamiltonian (such as one-body, or three-body, etc.), finding that the general principles derived for the sudden quench with two-body interactions apply universally. As an example, this is demonstrated for a conventional volumetric Otto cycle, where beneficial net work is generated by leveraging the control over the frequency of an external trap, which is a one-body operator. However, we emphasize that the results derived here apply universally to all Otto engine cycles operating under a sudden quench protocol. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:38:00 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Watson",
"R. S.",
""
],
[
"Kheruntsyan",
"K. V.",
""
]
] |
2505.21934 | Sarah Vollert | Sarah A. Vollert, Christopher Drovandi, Cailan Jeynes-Smith, Luz V. Pascal, Matthew P. Adams | Beyond data: leveraging non-empirical information and expert knowledge in Bayesian model calibration | null | null | null | null | stat.ME stat.AP | http://creativecommons.org/licenses/by/4.0/ | Mathematical models connect theory with the real world through data, enabling us to interpret, understand, and predict complex phenomena. However, scientific knowledge often extends beyond what can be empirically measured, offering valuable insights into complex and uncertain systems. Here, we introduce a statistical framework for calibrating mathematical models using non-empirical information. Through examples in ecology, biology, and medicine, we demonstrate how expert knowledge, scientific theory, and qualitative observations can meaningfully constrain models. In each case, these non-empirical insights guide models toward more realistic dynamics and more informed predictions than empirical data alone could achieve. Now, our understanding of the systems represented by mathematical models is not limited by the data that can be obtained; they instead sit at the edge of scientific understanding. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:38:59 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Vollert",
"Sarah A.",
""
],
[
"Drovandi",
"Christopher",
""
],
[
"Jeynes-Smith",
"Cailan",
""
],
[
"Pascal",
"Luz V.",
""
],
[
"Adams",
"Matthew P.",
""
]
] |
2505.21935 | Kaiyu He | Kaiyu He and Zhiyu Chen | From Reasoning to Learning: A Survey on Hypothesis Discovery and Rule Learning with Large Language Models | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Since the advent of Large Language Models (LLMs), efforts have largely focused on improving their instruction-following and deductive reasoning abilities, leaving open the question of whether these models can truly discover new knowledge. In pursuit of artificial general intelligence (AGI), there is a growing need for models that not only execute commands or retrieve information but also learn, reason, and generate new knowledge by formulating novel hypotheses and theories that deepen our understanding of the world. Guided by Peirce's framework of abduction, deduction, and induction, this survey offers a structured lens to examine LLM-based hypothesis discovery. We synthesize existing work in hypothesis generation, application, and validation, identifying both key achievements and critical gaps. By unifying these threads, we illuminate how LLMs might evolve from mere ``information executors'' into engines of genuine innovation, potentially transforming research, science, and real-world problem solving. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:40:02 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"He",
"Kaiyu",
""
],
[
"Chen",
"Zhiyu",
""
]
] |
2505.21936 | Zeyi Liao | Zeyi Liao, Jaylen Jones, Linxi Jiang, Eric Fosler-Lussier, Yu Su, Zhiqiang Lin, Huan Sun | RedTeamCUA: Realistic Adversarial Testing of Computer-Use Agents in Hybrid Web-OS Environments | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Computer-use agents (CUAs) promise to automate complex tasks across operating systems (OS) and the web, but remain vulnerable to indirect prompt injection. Current evaluations of this threat either lack support realistic but controlled environments or ignore hybrid web-OS attack scenarios involving both interfaces. To address this, we propose RedTeamCUA, an adversarial testing framework featuring a novel hybrid sandbox that integrates a VM-based OS environment with Docker-based web platforms. Our sandbox supports key features tailored for red teaming, such as flexible adversarial scenario configuration, and a setting that decouples adversarial evaluation from navigational limitations of CUAs by initializing tests directly at the point of an adversarial injection. Using RedTeamCUA, we develop RTC-Bench, a comprehensive benchmark with 864 examples that investigate realistic, hybrid web-OS attack scenarios and fundamental security vulnerabilities. Benchmarking current frontier CUAs identifies significant vulnerabilities: Claude 3.7 Sonnet | CUA demonstrates an ASR of 42.9%, while Operator, the most secure CUA evaluated, still exhibits an ASR of 7.6%. Notably, CUAs often attempt to execute adversarial tasks with an Attempt Rate as high as 92.5%, although failing to complete them due to capability limitations. Nevertheless, we observe concerning ASRs of up to 50% in realistic end-to-end settings, with the recently released frontier Claude 4 Opus | CUA showing an alarming ASR of 48%, demonstrating that indirect prompt injection presents tangible risks for even advanced CUAs despite their capabilities and safeguards. Overall, RedTeamCUA provides an essential framework for advancing realistic, controlled, and systematic analysis of CUA vulnerabilities, highlighting the urgent need for robust defenses to indirect prompt injection prior to real-world deployment. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:42:09 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Liao",
"Zeyi",
""
],
[
"Jones",
"Jaylen",
""
],
[
"Jiang",
"Linxi",
""
],
[
"Fosler-Lussier",
"Eric",
""
],
[
"Su",
"Yu",
""
],
[
"Lin",
"Zhiqiang",
""
],
[
"Sun",
"Huan",
""
]
] |
2505.21937 | Mohammadi Zaki | Pratik Rakesh Singh, Kritarth Prasad, Mohammadi Zaki and Pankaj Wasnik | Graph-Assisted Culturally Adaptable Idiomatic Translation for Indic Languages | null | ACL Findings 2025 | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Translating multi-word expressions (MWEs) and idioms requires a deep understanding of the cultural nuances of both the source and target languages. This challenge is further amplified by the one-to-many nature of idiomatic translations, where a single source idiom can have multiple target-language equivalents depending on cultural references and contextual variations. Traditional static knowledge graphs (KGs) and prompt-based approaches struggle to capture these complex relationships, often leading to suboptimal translations. To address this, we propose IdiomCE, an adaptive graph neural network (GNN) based methodology that learns intricate mappings between idiomatic expressions, effectively generalizing to both seen and unseen nodes during training. Our proposed method enhances translation quality even in resource-constrained settings, facilitating improved idiomatic translation in smaller models. We evaluate our approach on multiple idiomatic translation datasets using reference-less metrics, demonstrating significant improvements in translating idioms from English to various Indian languages. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:42:16 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Singh",
"Pratik Rakesh",
""
],
[
"Prasad",
"Kritarth",
""
],
[
"Zaki",
"Mohammadi",
""
],
[
"Wasnik",
"Pankaj",
""
]
] |
2505.21938 | Qirun Zeng | Qirun Zeng, Eric He, Richard Hoffmann, Xuchuang Wang, Jinhang Zuo | Practical Adversarial Attacks on Stochastic Bandits via Fake Data Injection | null | null | null | null | cs.LG cs.AI cs.CR | http://creativecommons.org/licenses/by/4.0/ | Adversarial attacks on stochastic bandits have traditionally relied on some unrealistic assumptions, such as per-round reward manipulation and unbounded perturbations, limiting their relevance to real-world systems. We propose a more practical threat model, Fake Data Injection, which reflects realistic adversarial constraints: the attacker can inject only a limited number of bounded fake feedback samples into the learner's history, simulating legitimate interactions. We design efficient attack strategies under this model, explicitly addressing both magnitude constraints (on reward values) and temporal constraints (on when and how often data can be injected). Our theoretical analysis shows that these attacks can mislead both Upper Confidence Bound (UCB) and Thompson Sampling algorithms into selecting a target arm in nearly all rounds while incurring only sublinear attack cost. Experiments on synthetic and real-world datasets validate the effectiveness of our strategies, revealing significant vulnerabilities in widely used stochastic bandit algorithms under practical adversarial scenarios. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:47:13 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zeng",
"Qirun",
""
],
[
"He",
"Eric",
""
],
[
"Hoffmann",
"Richard",
""
],
[
"Wang",
"Xuchuang",
""
],
[
"Zuo",
"Jinhang",
""
]
] |
2505.21939 | Dahoon Lee | Dahoon Lee, Chenglin Fan, Euiwoong Lee | Improved Approximation Algorithms for Chromatic and Pseudometric-Weighted Correlation Clustering | null | null | null | null | cs.DS | http://creativecommons.org/licenses/by/4.0/ | Correlation Clustering (CC) is a foundational problem in unsupervised learning that models binary similarity relations using labeled graphs. While classical CC has been widely studied, many real-world applications involve more nuanced relationships, either multi-class categorical interactions or varying confidence levels in edge labels. To address these, two natural generalizations have been proposed: Chromatic Correlation Clustering (CCC), which assigns semantic colors to edge labels, and pseudometric-weighted CC, which allows edge weights satisfying the triangle inequality. In this paper, we develop improved approximation algorithms for both settings. Our approach leverages LP-based pivoting techniques combined with problem-specific rounding functions. For the pseudometric-weighted correlation clustering problem, we present a tight $10/3$-approximation algorithm, matching the best possible bound achievable within the framework of standard LP relaxation combined with specialized rounding. For the Chromatic Correlation Clustering (CCC) problem, we improve the approximation ratio from the previous best of $2.5$ to $2.15$, and we establish a lower bound of $2.11$ within the same analytical framework, highlighting the near-optimality of our result. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:47:42 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lee",
"Dahoon",
""
],
[
"Fan",
"Chenglin",
""
],
[
"Lee",
"Euiwoong",
""
]
] |
2505.21940 | Xinran He | Bolei He, Xinran He, Mengke Chen, Xianwei Xue, Ying Zhu, Zhenhua Ling | RISE: Reasoning Enhancement via Iterative Self-Exploration in Multi-hop Question Answering | ACL 2025 Findings | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large Language Models (LLMs) excel in many areas but continue to face challenges with complex reasoning tasks, such as Multi-Hop Question Answering (MHQA). MHQA requires integrating evidence from diverse sources while managing intricate logical dependencies, often leads to errors in reasoning. Retrieval-Augmented Generation (RAG), widely employed in MHQA tasks, faces challenges in effectively filtering noisy data and retrieving all necessary evidence, thereby limiting its effectiveness in addressing MHQA challenges. To address these challenges, we propose RISE:Reasoning Enhancement via Iterative Self-Exploration, a novel framework designed to enhance models' reasoning capability through iterative self-exploration. Specifically, RISE involves three key steps in addressing MHQA tasks: question decomposition, retrieve-then-read, and self-critique. By leveraging continuous self-exploration, RISE identifies accurate reasoning paths, iteratively self-improving the model's capability to integrate evidence, maintain logical consistency, and enhance performance in MHQA tasks. Extensive experiments on multiple MHQA benchmarks demonstrate that RISE significantly improves reasoning accuracy and task performance. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:48:28 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"He",
"Bolei",
""
],
[
"He",
"Xinran",
""
],
[
"Chen",
"Mengke",
""
],
[
"Xue",
"Xianwei",
""
],
[
"Zhu",
"Ying",
""
],
[
"Ling",
"Zhenhua",
""
]
] |
2505.21941 | Ashim Gupta | Ashim Gupta, Vivek Srikumar | Test-Time Scaling with Repeated Sampling Improves Multilingual Text Generation | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Inference-time scaling via repeated sampling has shown promise in reasoning tasks, but its effectiveness in multilingual generation remains underexplored. We evaluate this approach using perplexity- and reward-based verifiers on two multilingual benchmarks: the Aya Evaluation Suite and m-ArenaHard. Our results show consistent quality improvements, with gains exceeding 35% in some cases. While perplexity-based scoring is effective for open-ended prompts, only reward-based verifiers improve performance on tasks requiring reasoning (e.g., math, code). Our results demonstrate the broader utility of repeated sampling for multilingual text generation and underscore the importance of selecting right verifiers for the task. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:50:19 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Gupta",
"Ashim",
""
],
[
"Srikumar",
"Vivek",
""
]
] |
2505.21942 | Prashant Bhat Mr | Prashant Bhat, Laurens Niesten, Elahe Arani, Bahram Zonooz | Continual Learning Beyond Experience Rehearsal and Full Model Surrogates | 23 pages, 9 figures | null | null | null | cs.LG stat.ML | http://creativecommons.org/licenses/by/4.0/ | Continual learning (CL) has remained a significant challenge for deep neural networks as learning new tasks erases previously acquired knowledge, either partially or completely. Existing solutions often rely on experience rehearsal or full model surrogates to mitigate CF. While effective, these approaches introduce substantial memory and computational overhead, limiting their scalability and applicability in real-world scenarios. To address this, we propose SPARC, a scalable CL approach that eliminates the need for experience rehearsal and full-model surrogates. By effectively combining task-specific working memories and task-agnostic semantic memory for cross-task knowledge consolidation, SPARC results in a remarkable parameter efficiency, using only 6% of the parameters required by full-model surrogates. Despite its lightweight design, SPARC achieves superior performance on Seq-TinyImageNet and matches rehearsal-based methods on various CL benchmarks. Additionally, weight re-normalization in the classification layer mitigates task-specific biases, establishing SPARC as a practical and scalable solution for CL under stringent efficiency constraints. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:52:34 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bhat",
"Prashant",
""
],
[
"Niesten",
"Laurens",
""
],
[
"Arani",
"Elahe",
""
],
[
"Zonooz",
"Bahram",
""
]
] |
2505.21943 | Wei Lin | Wei Lin, Chenyang Zhao, and Antoni B. Chan | Point-to-Region Loss for Semi-Supervised Point-Based Crowd Counting | accepted by CVPR-2025(highlight) | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Point detection has been developed to locate pedestrians in crowded scenes by training a counter through a point-to-point (P2P) supervision scheme. Despite its excellent localization and counting performance, training a point-based counter still faces challenges concerning annotation labor: hundreds to thousands of points are required to annotate a single sample capturing a dense crowd. In this paper, we integrate point-based methods into a semi-supervised counting framework based on pseudo-labeling, enabling the training of a counter with only a few annotated samples supplemented by a large volume of pseudo-labeled data. However, during implementation, the training encounters issues as the confidence for pseudo-labels fails to be propagated to background pixels via the P2P. To tackle this challenge, we devise a point-specific activation map (PSAM) to visually interpret the phenomena occurring during the ill-posed training. Observations from the PSAM suggest that the feature map is excessively activated by the loss for unlabeled data, causing the decoder to misinterpret these over-activations as pedestrians. To mitigate this issue, we propose a point-to-region (P2R) scheme to substitute P2P, which segments out local regions rather than detects a point corresponding to a pedestrian for supervision. Consequently, pixels in the local region can share the same confidence with the corresponding pseudo points. Experimental results in both semi-supervised counting and unsupervised domain adaptation highlight the advantages of our method, illustrating P2R can resolve issues identified in PSAM. The code is available at https://github.com/Elin24/P2RLoss. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:53:08 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lin",
"Wei",
""
],
[
"Zhao",
"Chenyang",
""
],
[
"Chan",
"Antoni B.",
""
]
] |
2505.21944 | Linli Zhou | Linli Zhou, Bokun Wang, My T. Thai, Tianbao Yang | Stochastic Primal-Dual Double Block-Coordinate for Two-way Partial AUC Maximization | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Two-way partial AUC (TPAUC) is a critical performance metric for binary classification with imbalanced data, as it focuses on specific ranges of the true positive rate (TPR) and false positive rate (FPR). However, stochastic algorithms for TPAUC optimization remain under-explored, with existing methods either limited to approximated TPAUC loss functions or burdened by sub-optimal complexities. To overcome these limitations, we introduce two innovative stochastic primal-dual double block-coordinate algorithms for TPAUC maximization. These algorithms utilize stochastic block-coordinate updates for both the primal and dual variables, catering to both convex and non-convex settings. We provide theoretical convergence rate analyses, demonstrating significant improvements over prior approaches. Our experimental results, based on multiple benchmark datasets, validate the superior performance of our algorithms, showcasing faster convergence and better generalization. This work advances the state of the art in TPAUC optimization and offers practical tools for real-world machine learning applications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:55:05 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhou",
"Linli",
""
],
[
"Wang",
"Bokun",
""
],
[
"Thai",
"My T.",
""
],
[
"Yang",
"Tianbao",
""
]
] |
2505.21945 | Maria Kirsanova S. | Maria. S. Kirsanova, Anastasiia A. Farafontova | Deuterated water abundance in the young hot core RCW 120 S2 | Submitted to Astronomy Letters, comments are welcome | null | null | null | astro-ph.SR astro-ph.GA | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Since the emission of water molecules cannot be observed from Earth, less abundant isotopologues, such as H$_2^{18}$O and HDO, are used to trace water in star-forming regions. The main aim of this study is to determine HDO abundance in the hot core RCW 120 S2. We performed observations of the hot core in the 200-255~GHz range using the nFLASH230 receiver on the APEX telescope. Two HDO lines were detected toward RCW 120 S2. Their intensities are described by excitation temperature $286\pm2$~K and gas number density $\geq 10^9$~cm$^{-3}$. The emission originates from the hot core rather than the warm dense envelope surrounding a central young stellar object. The HDO column density ranges from $(3.9-7.9)\times10^{13}$~cm$^{-3}$ with a most probable value $5.6\times10^{13}$~cm$^{-3}$. The HDO abundance relative to hydrogen nuclei is $1.5\times10^{-9}$. This HDO abundance value is among the lowest reported for hot cores. Combined with the non-detection of the H$_2^{18}$O line, we conclude that protostellar heating in RCW 120 S2 is still in its early stages. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:55:13 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Kirsanova",
"Maria. S.",
""
],
[
"Farafontova",
"Anastasiia A.",
""
]
] |
2505.21946 | Sinan Wang | Sinan Wang, Junwei Zhou, Fan Feng, Zhiqi Li, Yuchen Sun, Duowen Chen, Greg Turk, Bo Zhu | Fluid Simulation on Vortex Particle Flow Maps | ACM Transactions on Graphics (SIGGRAPH 2025), 24 pages | null | 10.1145/3731198 | null | cs.GR physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | We propose the Vortex Particle Flow Map (VPFM) method to simulate incompressible flow with complex vortical evolution in the presence of dynamic solid boundaries. The core insight of our approach is that vorticity is an ideal quantity for evolution on particle flow maps, enabling significantly longer flow map distances compared to other fluid quantities like velocity or impulse. To achieve this goal, we developed a hybrid Eulerian-Lagrangian representation that evolves vorticity and flow map quantities on vortex particles, while reconstructing velocity on a background grid. The method integrates three key components: (1) a vorticity-based particle flow map framework, (2) an accurate Hessian evolution scheme on particles, and (3) a solid boundary treatment for no-through and no-slip conditions in VPFM. These components collectively allow a substantially longer flow map length (3-12 times longer) than the state-of-the-art, enhancing vorticity preservation over extended spatiotemporal domains. We validated the performance of VPFM through diverse simulations, demonstrating its effectiveness in capturing complex vortex dynamics and turbulence phenomena. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:56:38 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wang",
"Sinan",
""
],
[
"Zhou",
"Junwei",
""
],
[
"Feng",
"Fan",
""
],
[
"Li",
"Zhiqi",
""
],
[
"Sun",
"Yuchen",
""
],
[
"Chen",
"Duowen",
""
],
[
"Turk",
"Greg",
""
],
[
"Zhu",
"Bo",
""
]
] |
2505.21947 | Luke Jonathan Mark Davies Dr | L. J. M. Davies, J. E. Thorne, S. Bellstedt, R. H. W. Cook, M. Bravo, A. S. G. Robotham, C. del P. Lagos, S. Phillipps, M. Siudek, B. W. Holwerda, M. N. Bremer, J. D'Silva and S. P. Driver | Deep Extragalactic VIsible Legacy Survey (DEVILS): The sSFR-M$_{\star}$plane part I: The recent SFH of galaxies and movement through the plane | 30 pages, 20 figures - Accepted MNRAS | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In a recent paper we parameterised the evolution of the star-formation rate dispersion ($\sigma_{SFR}$) across the specific star-formation rate - stellar mass plane (sSFR-M$_{\star}$) using the Deep Extragalactic VIsible Legacy Survey (DEVILS) - suggesting that the point at which the minimum in the dispersion occurs (M$^{*}_{\sigma-min}$) defines a boundary between different physical mechanisms affecting galaxy evolution. Here we expand upon that work to determine the movement of galaxies through the sSFR-M$_{\star}$ plane using their recent star-formation histories (SFHs) and explore how this leads to the observed $\sigma_{SFR}$-M$_{\star}$ relation. We find that galaxies in sub-regions of the sSFR-M$_{\star}$ plane show distinctly different SFHs, leading to a complex evolution of the sSFR-M$_{\star}$ plane and star-forming sequence (SFS). However, we find that selecting galaxies based on stellar mass and position relative to SFS alone (as is traditionally the case), may not identify sources with common recent SFHs, and therefore propose a new selection methodology. We then use the recent SFH of galaxies to measure the evolution of the SFS, showing that it has varying contributions from galaxies with different SFHs that lead to the observed changes in slope, normalisation and turnover stellar mass. Finally, we determine the overall evolution of the sSFR-M$_{\star}$ plane from $z\sim1$ to today. In the second paper in this series we will discuss physical properties of galaxies with common recent SFHs and how these lead to the observed $\sigma_{SFR}$-M$_{\star}$ relation and evolution of the sSFR-M$_{\star}$ plane. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:56:57 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Davies",
"L. J. M.",
""
],
[
"Thorne",
"J. E.",
""
],
[
"Bellstedt",
"S.",
""
],
[
"Cook",
"R. H. W.",
""
],
[
"Bravo",
"M.",
""
],
[
"Robotham",
"A. S. G.",
""
],
[
"Lagos",
"C. del P.",
""
],
[
"Phillipps",
"S.",
""
],
[
"Siudek",
"M.",
""
],
[
"Holwerda",
"B. W.",
""
],
[
"Bremer",
"M. N.",
""
],
[
"D'Silva",
"J.",
""
],
[
"Driver",
"S. P.",
""
]
] |
2505.21948 | Luke Jonathan Mark Davies Dr | L. J. M. Davies, J. E. Thorne, S. Bellstedt, R. H. W. Cook, M. Bravo, A. S. G. Robotham, C. del P. Lagos, S. Phillipps, M. Siudek, B. W. Holwerda, M. N. Bremer, J. D'Silva, and S. P. Driver | Deep Extragalactic VIsible Legacy Survey (DEVILS): The sSFR-M$_{\star}$ plane part II: Starbursts, SFHs and AGN Feedback | 31 pages, 14 figures, Accepted - MNRAS | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In part I of this series we discussed the variation of star-formation histories (SFHs) across the specific star formation rate - stellar mass plane (sSFR-M$_{\star}$) using the Deep Extragalactic VIsible Legacy Survey (DEVILS). Here we explore the physical mechanisms that are likely driving these observational trends, by comparing the properties of galaxies with common recent SFH shapes. Overall, we find that the processes shaping the movement of galaxies through the sSFR-M$_{\star}$ plane can be be largely split into two stellar mass regimes, bounded by the minimum SFR dispersion ($\sigma_{SFR}$) point. At lower stellar masses we find that large $\sigma_{SFR}$ values are likely observed due to a combination of stochastic star-formation processes and a large variety in absolute sSFR values, but relatively constant/flat SFHs. While at higher stellar masses we see strong observational evidence that Active Galactic Nuclei (AGN) are associated with rapidly declining SFHs, and that these galaxies reside in the high $\sigma_{SFR}$ region of the plane. As such, we suggest that AGN feedback, leading to galaxy quenching, is the primary driver of the high $\sigma_{SFR}$ values. These results are consistent with previous theoretical interpretations of the $\sigma_{SFR}$-M$_{\star}$ relation. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:57:04 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Davies",
"L. J. M.",
""
],
[
"Thorne",
"J. E.",
""
],
[
"Bellstedt",
"S.",
""
],
[
"Cook",
"R. H. W.",
""
],
[
"Bravo",
"M.",
""
],
[
"Robotham",
"A. S. G.",
""
],
[
"Lagos",
"C. del P.",
""
],
[
"Phillipps",
"S.",
""
],
[
"Siudek",
"M.",
""
],
[
"Holwerda",
"B. W.",
""
],
[
"Bremer",
"M. N.",
""
],
[
"D'Silva",
"J.",
""
],
[
"Driver",
"S. P.",
""
]
] |
2505.21949 | Jiayi Liu | Jiayi Liu, Jonathan E. Ron, Giulia Rinaldi, Ivanna Williantarra, Antonios Georgantzoglou, Ingrid de Vries, Michael Sixt, Milka Sarris, Nir S. Gov | Chemotaxis of branched cells in complex environments | null | null | null | null | physics.bio-ph q-bio.CB | http://creativecommons.org/licenses/by/4.0/ | Cell migration in vivo is often guided by chemical signals. Such chemotaxis, such as performed by immune cells migrating to a wound site, is complicated by the complex geometry inside living tissues. In this study, we extend our theoretical model of branched-cell migration on a network by introducing chemokine sources to explore the cellular response. The model predicts a speed-accuracy tradeoff, whereby slow cells are significantly more accurate and able to follow efficiently a weak chemoattractant signal. We then compare the model's predictions with experimental observations of neutrophils migrating to the site of laser-inflicted wound in a zebrafish larva fin, and migrating in-vitro inside a regular lattice of pillars. We find that the model captures the details of the sub-cellular response to the chemokine gradient, as well as the large-scale migration response. This comparison suggests that the neutrophils behave as fast cells, compromising their chemotaxis accuracy, which explains the functionality of these immune cells. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:00:43 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Liu",
"Jiayi",
""
],
[
"Ron",
"Jonathan E.",
""
],
[
"Rinaldi",
"Giulia",
""
],
[
"Williantarra",
"Ivanna",
""
],
[
"Georgantzoglou",
"Antonios",
""
],
[
"de Vries",
"Ingrid",
""
],
[
"Sixt",
"Michael",
""
],
[
"Sarris",
"Milka",
""
],
[
"Gov",
"Nir S.",
""
]
] |
2505.21950 | Joshua Rooney | Joshua Rooney | Winning Probabilities of Balanced and Nontransitive n-tuples of Dice | null | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | For a positive integer $n$, an $n$-tuple of dice $(A_1,A_2,\dots,A_n)$ is called balanced if $P(A_1<A_2) = P(A_2<A_3) = \cdots = P(A_n<A_1)$ and nontransitive if $P(A_1<A_2), P(A_2<A_3), \dots, P(A_n<A_1)$ are each greater than $\frac{1}{2}$. For a balanced and nontransitive $n$-tuple of dice $(A_1,A_2,\dots,A_n)$, we define the winning probability $w(A_1,A_2,\dots,A_n) := P(A_1 < A_2)$. The works of Trybula and Kim et al. together show that for a balanced and nontransitve triple of dice $(A_1,A_2,A_3)$, the least upper bound on the winning probability is $\frac{-1+\sqrt{5}}{2}$. Kim et al. then asked what the least upper bound on the winning probability was for the $n \geq 4$ cases. Bogdanov and Komisarski independently have shown that for $n\geq 3$ and a balanced and nontransitive $n$-tuple of dice $(A_1,A_2,\dots,A_n)$, the winning probability is less than $\pi_n := 1-\frac{1}{4\cos^2\left( \frac{\pi}{n+2} \right)}$.
In this paper, we will show that for $n \geq 3$ and every rational $p \in \left( \frac{1}{2}, \pi_n \right]$, there exists a balanced and nontransitive $n$-tuple of dice with winning probability $p$. Paired with Bogdanov and Komisarski's results, this fully answers the problem posed by Kim et al. and establishes a complete characterization of the winning probabilities for nontransitive and balanced $n$-tuples of dice. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:03:03 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Rooney",
"Joshua",
""
]
] |
2505.21951 | Yulin Shao | Zijian Yang, Yulin Shao, Shaodan Ma | When Feedback Empowers the Uplink: Integrating Adaptive Coding with Wireless Power Transfer | null | null | null | null | cs.IT eess.SP math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Energy consumption and device lifetime are critical concerns for battery-constrained IoT devices. This paper introduces the Feedback-Aided Coding and Energy Transfer (FACET) framework, which synergistically combines adaptive feedback channel coding with wireless power transfer. FACET leverages the saturation effect of feedback coding, where increasing downlink power yields diminishing returns, to design a dual-purpose feedback mechanism that simultaneously guides uplink coding and replenishes device energy. We characterize the inherent tradeoff between feedback precision and harvested power, and formulate a fairness-constrained min-max optimization problem to minimize worst-case net energy consumption. An efficient algorithm based on alternating optimization and Lagrangian duality is developed, with each subproblem admitting a closed-form solution. Simulations show that FACET nearly triples device lifetime compared to conventional feedback coding architectures, and remains robust across a wide range of power regimes. These results suggest that FACET not only improves communication efficiency but also redefines the role of feedback in energy-constrained IoT systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:04:24 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Yang",
"Zijian",
""
],
[
"Shao",
"Yulin",
""
],
[
"Ma",
"Shaodan",
""
]
] |
2505.21952 | Masahiko Ueda | Masahiko Ueda | Properties of zero-determinant strategies in multichannel games | 12 pages | null | null | null | physics.soc-ph cs.GT cs.MA cs.SY eess.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Controlling payoffs in repeated games is one of the important topics in control theory of multi-agent systems. Recently proposed zero-determinant strategies enable players to unilaterally enforce linear relations between payoffs. Furthermore, based on the mathematics of zero-determinant strategies, regional payoff control, in which payoffs are enforced into some feasible regions, has been discovered in social dilemma situations. More recently, theory of payoff control was extended to multichannel games, where players parallelly interact with each other in multiple channels. However, properties of zero-determinant strategies specific to multichannel games are still not clear. In this paper, we elucidate properties of zero-determinant strategies in multichannel games. First, we relate the existence condition of zero-determinant strategies in multichannel games to that of zero-determinant strategies in each channel. We then show that the existence of zero-determinant strategies in multichannel games requires the existence of zero-determinant strategies in some channels. This result implies that the existence of zero-determinant strategies in multichannel games is tightly restricted by structure of games played in each channel. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:06:04 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ueda",
"Masahiko",
""
]
] |
2505.21953 | Kavita Kumari | Kavita Kumari, I. E. Papadakis and G. C. Dewangan | Exploring the Accretion disc/Corona Connection in NGC 6814: Insights from UV and X-ray spectral-timing studies | Accepted for publication in ApJ, 20 Pages, 15 figures | null | null | null | astro-ph.HE astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We conducted a comprehensive spectral and timing analysis of NGC 6814 using AstroSat's 2019 and XMM-Newton's 2021 observations. Cross-correlation analysis revealed a significant correlation between FUV (1541 \AA)/X-ray and UVW1 (2910 \AA)/X-ray variations, with delays of $\sim 15~\rm{ks}$ and $30~\rm{ks}$, respectively. We constructed four broadband SEDs after applying aperture correction (for the UVIT filter), subtracting host galaxy and emission line contributions from UV flux, and using mean X-ray spectra alongside selected UV data points. First, we fitted the SEDs with KYNSED model assuming various combinations of inclination, $\theta$, color correction factors, $f_{\rm col}$, and BH spins. Best-fit models were achieved for $\theta=70^{\circ}$ (consistent with past estimates for this source) and for spin $\leq 0.5$, while $f_{\rm col}$ is not constrained. KYNSED provided satisfactory fit to all SEDs in the case when the corona is powered by the accretion process, with $\sim 10-20$% of the accretion power transferred to the corona, $\dot{m}/\dot{m}_{\rm Edd}\sim 0.1$, corona radius of $\sim 6-10~r_g$, and height of $\sim7.5-35~r_g$. Model time-lags computed using the SED best-fit results are aligned well with the observed time-lags. Although some of the model parameters are not constrained, the important result of our work is that both the broadband X-ray/UV spectra and the X-ray/UV time-lags in NGC 6814 are consistent with the hypothesis of X-ray illumination of the disc in a lamp-post geometry framework. Within this model framework, we do not need to assume an outer or inner truncated disc. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:07:49 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Kumari",
"Kavita",
""
],
[
"Papadakis",
"I. E.",
""
],
[
"Dewangan",
"G. C.",
""
]
] |
2505.21954 | Le Thien Phuc Nguyen | Le Thien Phuc Nguyen, Zhuoran Yu, Khoa Quang Nhat Cao, Yuwei Guo, Tu Ho Manh Pham, Tuan Tai Nguyen, Toan Ngo Duc Vo, Lucas Poon, Soochahn Lee, Yong Jae Lee | UniTalk: Towards Universal Active Speaker Detection in Real World Scenarios | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | We present UniTalk, a novel dataset specifically designed for the task of active speaker detection, emphasizing challenging scenarios to enhance model generalization. Unlike previously established benchmarks such as AVA, which predominantly features old movies and thus exhibits significant domain gaps, UniTalk focuses explicitly on diverse and difficult real-world conditions. These include underrepresented languages, noisy backgrounds, and crowded scenes - such as multiple visible speakers speaking concurrently or in overlapping turns. It contains over 44.5 hours of video with frame-level active speaker annotations across 48,693 speaking identities, and spans a broad range of video types that reflect real-world conditions. Through rigorous evaluation, we show that state-of-the-art models, while achieving nearly perfect scores on AVA, fail to reach saturation on UniTalk, suggesting that the ASD task remains far from solved under realistic conditions. Nevertheless, models trained on UniTalk demonstrate stronger generalization to modern "in-the-wild" datasets like Talkies and ASW, as well as to AVA. UniTalk thus establishes a new benchmark for active speaker detection, providing researchers with a valuable resource for developing and evaluating versatile and resilient models.
Dataset: https://huggingface.co/datasets/plnguyen2908/UniTalk-ASD
Code: https://github.com/plnguyen2908/UniTalk-ASD-code | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:08:59 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Nguyen",
"Le Thien Phuc",
""
],
[
"Yu",
"Zhuoran",
""
],
[
"Cao",
"Khoa Quang Nhat",
""
],
[
"Guo",
"Yuwei",
""
],
[
"Pham",
"Tu Ho Manh",
""
],
[
"Nguyen",
"Tuan Tai",
""
],
[
"Vo",
"Toan Ngo Duc",
""
],
[
"Poon",
"Lucas",
""
],
[
"Lee",
"Soochahn",
""
],
[
"Lee",
"Yong Jae",
""
]
] |
2505.21955 | Insu Lee | Insu Lee, Wooje Park, Jaeyun Jang, Minyoung Noh, Kyuhong Shim, Byonghyo Shim | Towards Comprehensive Scene Understanding: Integrating First and Third-Person Views for LVLMs | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large vision-language models (LVLMs) are increasingly deployed in interactive applications such as virtual and augmented reality, where first-person (egocentric) view captured by head-mounted cameras serves as key input. While this view offers fine-grained cues about user attention and hand-object interactions, their narrow field of view and lack of global context often lead to failures on spatially or contextually demanding queries. To address this, we introduce a framework that augments egocentric inputs with third-person (exocentric) views, providing complementary information such as global scene layout and object visibility to LVLMs. We present E3VQA, the first benchmark for multi-view question answering with 4K high-quality question-answer pairs grounded in synchronized ego-exo image pairs. Additionally, we propose M3CoT, a training-free prompting technique that constructs a unified scene representation by integrating scene graphs from three complementary perspectives. M3CoT enables LVLMs to reason more effectively across views, yielding consistent performance gains (4.84% for GPT-4o and 5.94% for Gemini 2.0 Flash) over a recent CoT baseline. Our extensive evaluation reveals key strengths and limitations of LVLMs in multi-view reasoning and highlights the value of leveraging both egocentric and exocentric inputs. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:09:42 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lee",
"Insu",
""
],
[
"Park",
"Wooje",
""
],
[
"Jang",
"Jaeyun",
""
],
[
"Noh",
"Minyoung",
""
],
[
"Shim",
"Kyuhong",
""
],
[
"Shim",
"Byonghyo",
""
]
] |
2505.21957 | Jinxiao Qin | Jinxiao Qin, Hong-Liang Yan, Wenyuan Cui, Jian-Rong Shi, Subo Dong, Shuai Liu, Zeming Zhou, Miao Tian, Zhenyan Huo, Xiangsong Fang, Jinghua Zhang, Chunqian Li, Mingyi Ding, Song Wang, and Henggeng Han | Revisiting the Li abundances of Stars with and without Detected Planets from the High Resolution Spectroscopy | 13 pages, 9 figures | null | null | null | astro-ph.SR astro-ph.EP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Whether the presence of planets affects the lithium (Li) abundance of their host stars is still an open question. To investigate the difference of the Li abundance between planet-host stars (HS) and isolated stars (IS) with no detected planets, we analyze a large sample of stars with temperatures ranging from 4600 to 6600 K and metallicity ranging from -0.55 to +0.50. The sample consists of 279 HS whose spectra were taken from the California-Kepler Survey (CKS), which followed up planets detected by Kepler, and 171 IS whose spectra were taken from the Keck archive. The non-local thermodynamic equilibrium (non-LTE) effects were taken into consideration. It is found that the distribution of Li abundances in both the HS and IS groups are generally consistent with each other. This suggests that the presence of Kepler-like planets does not have a significant impact on Li depletion. We also found that the non-LTE corrections can not be neglected for stars with A(Li) over ~ 2.5 dex. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:13:51 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Qin",
"Jinxiao",
""
],
[
"Yan",
"Hong-Liang",
""
],
[
"Cui",
"Wenyuan",
""
],
[
"Shi",
"Jian-Rong",
""
],
[
"Dong",
"Subo",
""
],
[
"Liu",
"Shuai",
""
],
[
"Zhou",
"Zeming",
""
],
[
"Tian",
"Miao",
""
],
[
"Huo",
"Zhenyan",
""
],
[
"Fang",
"Xiangsong",
""
],
[
"Zhang",
"Jinghua",
""
],
[
"Li",
"Chunqian",
""
],
[
"Ding",
"Mingyi",
""
],
[
"Wang",
"Song",
""
],
[
"Han",
"Henggeng",
""
]
] |
2505.21958 | Qihuang Zhong | Qihuang Zhong, Liang Ding, Fei Liao, Juhua Liu, Bo Du, Dacheng Tao | Resolving Knowledge Conflicts in Domain-specific Data Selection: A Case Study on Medical Instruction-tuning | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Domain-specific instruction-tuning has become the defacto standard for improving the performance of large language models (LLMs) in specialized applications, e.g., medical question answering. Since the instruction-tuning dataset might contain redundant or low-quality data, data selection (DS) is usually required to maximize the data efficiency. Despite the successes in the general domain, current DS methods often struggle to select the desired data for domain-specific instruction-tuning. One of the main reasons is that they neglect the impact of knowledge conflicts, i.e., the discrepancy between LLMs' pretrained knowledge and context knowledge of instruction data, which could damage LLMs' prior abilities and lead to hallucination. To this end, we propose a simple-yet-effective Knowledge-aware Data Selection (namely KDS) framework to select the domain-specific instruction-tuning data that meets LLMs' actual needs. The core of KDS is to leverage two knowledge-aware metrics for quantitatively measuring knowledge conflicts from two aspects: context-memory knowledge alignment and intra-memory knowledge consistency. By filtering the data with large knowledge conflicts and sampling the high-quality and diverse data, KDS can effectively stimulate the LLMs' abilities and achieve better domain-specific performance. Taking the medical domain as the testbed, we conduct extensive experiments and empirically prove that KDS surpasses the other baselines and brings significant and consistent performance gains among all LLMs. More encouragingly, KDS effectively improves the model generalization and alleviates the hallucination problem. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:18:24 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhong",
"Qihuang",
""
],
[
"Ding",
"Liang",
""
],
[
"Liao",
"Fei",
""
],
[
"Liu",
"Juhua",
""
],
[
"Du",
"Bo",
""
],
[
"Tao",
"Dacheng",
""
]
] |
2505.21959 | Aakriti Agrawal | Aakriti Agrawal, Mucong Ding, Zora Che, Chenghao Deng, Anirudh Satheesh, Bang An, Bayan Bruss, John Langford, Furong Huang | EnsemW2S: Enhancing Weak-to-Strong Generalization with Large Language Model Ensembles | Superalignment. arXiv admin note: substantial text overlap with arXiv:2410.04571 | null | null | null | cs.LG cs.CL | http://creativecommons.org/licenses/by/4.0/ | With Large Language Models (LLMs) rapidly approaching and potentially surpassing human-level performance, it has become imperative to develop approaches capable of effectively supervising and enhancing these powerful models using smaller, human-level models exposed to only human-level data. We address this critical weak-to-strong (W2S) generalization challenge by proposing a novel method aimed at improving weak experts, by training on the same limited human-level data, enabling them to generalize to complex, super-human-level tasks. Our approach, called \textbf{EnsemW2S}, employs a token-level ensemble strategy that iteratively combines multiple weak experts, systematically addressing the shortcomings identified in preceding iterations. By continuously refining these weak models, we significantly enhance their collective ability to supervise stronger student models. We extensively evaluate the generalization performance of both the ensemble of weak experts and the subsequent strong student model across in-distribution (ID) and out-of-distribution (OOD) datasets. For OOD, we specifically introduce question difficulty as an additional dimension for defining distributional shifts. Our empirical results demonstrate notable improvements, achieving 4\%, and 3.2\% improvements on ID datasets and, upto 6\% and 2.28\% on OOD datasets for experts and student models respectively, underscoring the effectiveness of our proposed method in advancing W2S generalization. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:23:12 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Agrawal",
"Aakriti",
""
],
[
"Ding",
"Mucong",
""
],
[
"Che",
"Zora",
""
],
[
"Deng",
"Chenghao",
""
],
[
"Satheesh",
"Anirudh",
""
],
[
"An",
"Bang",
""
],
[
"Bruss",
"Bayan",
""
],
[
"Langford",
"John",
""
],
[
"Huang",
"Furong",
""
]
] |
2505.21960 | Senmao Li | Senmao Li, Lei Wang, Kai Wang, Tao Liu, Jiehang Xie, Joost van de Weijer, Fahad Shahbaz Khan, Shiqi Yang, Yaxing Wang, Jian Yang | One-Way Ticket:Time-Independent Unified Encoder for Distilling Text-to-Image Diffusion Models | Accepted at CVPR2025, Code: https://github.com/sen-mao/Loopfree | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Text-to-Image (T2I) diffusion models have made remarkable advancements in generative modeling; however, they face a trade-off between inference speed and image quality, posing challenges for efficient deployment. Existing distilled T2I models can generate high-fidelity images with fewer sampling steps, but often struggle with diversity and quality, especially in one-step models. From our analysis, we observe redundant computations in the UNet encoders. Our findings suggest that, for T2I diffusion models, decoders are more adept at capturing richer and more explicit semantic information, while encoders can be effectively shared across decoders from diverse time steps. Based on these observations, we introduce the first Time-independent Unified Encoder TiUE for the student model UNet architecture, which is a loop-free image generation approach for distilling T2I diffusion models. Using a one-pass scheme, TiUE shares encoder features across multiple decoder time steps, enabling parallel sampling and significantly reducing inference time complexity. In addition, we incorporate a KL divergence term to regularize noise prediction, which enhances the perceptual realism and diversity of the generated images. Experimental results demonstrate that TiUE outperforms state-of-the-art methods, including LCM, SD-Turbo, and SwiftBrushv2, producing more diverse and realistic results while maintaining the computational efficiency. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:23:22 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Li",
"Senmao",
""
],
[
"Wang",
"Lei",
""
],
[
"Wang",
"Kai",
""
],
[
"Liu",
"Tao",
""
],
[
"Xie",
"Jiehang",
""
],
[
"van de Weijer",
"Joost",
""
],
[
"Khan",
"Fahad Shahbaz",
""
],
[
"Yang",
"Shiqi",
""
],
[
"Wang",
"Yaxing",
""
],
[
"Yang",
"Jian",
""
]
] |
2505.21961 | Vahid Mousavi | S. V. Mousavi | Tripartite Entanglement dynamics: the influence of intrinsic decoherence and decoherence channels | Accepted for publication in Quantum Information Processing | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study examines a system of three coupled qubits, focusing on entanglement measures in the presence of decoherence. It utilizes an XXZ Heisenberg chain with an external magnetic field and Dzyaloshinskii-Moriya interaction, considering intrinsic decoherence. The results reveal that only the magnetic field strength affects entanglement, while intrinsic decoherence suppresses it, with stronger decoherence leading to greater suppression. Various decoherence channels are analyzed, showing that the $I$-tangle typically decreases with increased decoherence, except for the generalized W state under phase damping channel, where only one qubit is affected. Interestingly, dark periods of $I$-tangle occur for the GHZ state under non-Markovian dephasing, and while steady-state entanglement disappears in this channel, it remains nonzero when starting from a mixture of GHZ and fully separable states. Additionally, under generalized amplitude damping channel, reduced bipartite states of a W state exhibit entanglement sudden death, while the steady-state $I$-tangle for the spectral decomposed state stays nonzero. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:23:31 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Mousavi",
"S. V.",
""
]
] |
2505.21962 | Mengjingcheng Mo | Mengjingcheng Mo, Xinyang Tong, Jiaxu Leng, Mingpi Tan, Jiankang Zheng, Yiran Liu, Haosheng Chen, Ji Gan, Weisheng Li, Xinbo Gao | A2Seek: Towards Reasoning-Centric Benchmark for Aerial Anomaly Understanding | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | While unmanned aerial vehicles (UAVs) offer wide-area, high-altitude coverage for anomaly detection, they face challenges such as dynamic viewpoints, scale variations, and complex scenes. Existing datasets and methods, mainly designed for fixed ground-level views, struggle to adapt to these conditions, leading to significant performance drops in drone-view scenarios. To bridge this gap, we introduce A2Seek (Aerial Anomaly Seek), a large-scale, reasoning-centric benchmark dataset for aerial anomaly understanding. This dataset covers various scenarios and environmental conditions, providing high-resolution real-world aerial videos with detailed annotations, including anomaly categories, frame-level timestamps, region-level bounding boxes, and natural language explanations for causal reasoning. Building on this dataset, we propose A2Seek-R1, a novel reasoning framework that generalizes R1-style strategies to aerial anomaly understanding, enabling a deeper understanding of "Where" anomalies occur and "Why" they happen in aerial frames. To this end, A2Seek-R1 first employs a graph-of-thought (GoT)-guided supervised fine-tuning approach to activate the model's latent reasoning capabilities on A2Seek. Then, we introduce Aerial Group Relative Policy Optimization (A-GRPO) to design rule-based reward functions tailored to aerial scenarios. Furthermore, we propose a novel "seeking" mechanism that simulates UAV flight behavior by directing the model's attention to informative regions. Extensive experiments demonstrate that A2Seek-R1 achieves up to a 22.04% improvement in AP for prediction accuracy and a 13.9% gain in mIoU for anomaly localization, exhibiting strong generalization across complex environments and out-of-distribution scenarios. Our dataset and code will be released at https://hayneyday.github.io/A2Seek/. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:28:13 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Mo",
"Mengjingcheng",
""
],
[
"Tong",
"Xinyang",
""
],
[
"Leng",
"Jiaxu",
""
],
[
"Tan",
"Mingpi",
""
],
[
"Zheng",
"Jiankang",
""
],
[
"Liu",
"Yiran",
""
],
[
"Chen",
"Haosheng",
""
],
[
"Gan",
"Ji",
""
],
[
"Li",
"Weisheng",
""
],
[
"Gao",
"Xinbo",
""
]
] |
2505.21963 | Taro Yano | Taro Yano, Yoichi Ishibashi, Masafumi Oyamada | LaMDAgent: An Autonomous Framework for Post-Training Pipeline Optimization via LLM Agents | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large Language Models (LLMs) have demonstrated exceptional performance across a wide range of tasks. To further tailor LLMs to specific domains or applications, post-training techniques such as Supervised Fine-Tuning (SFT), Preference Learning, and model merging are commonly employed. While each of these methods has been extensively studied in isolation, the automated construction of complete post-training pipelines remains an underexplored area. Existing approaches typically rely on manual design or focus narrowly on optimizing individual components, such as data ordering or merging strategies. In this work, we introduce LaMDAgent (short for Language Model Developing Agent), a novel framework that autonomously constructs and optimizes full post-training pipelines through the use of LLM-based agents. LaMDAgent systematically explores diverse model generation techniques, datasets, and hyperparameter configurations, leveraging task-based feedback to discover high-performing pipelines with minimal human intervention. Our experiments show that LaMDAgent improves tool-use accuracy by 9.0 points while preserving instruction-following capabilities. Moreover, it uncovers effective post-training strategies that are often overlooked by conventional human-driven exploration. We further analyze the impact of data and model size scaling to reduce computational costs on the exploration, finding that model size scalings introduces new challenges, whereas scaling data size enables cost-effective pipeline discovery. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:30:51 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Yano",
"Taro",
""
],
[
"Ishibashi",
"Yoichi",
""
],
[
"Oyamada",
"Masafumi",
""
]
] |
2505.21964 | Xiaoyi Zhang | Ziyun Zhang, Xinyi Liu, Xiaoyi Zhang, Jun Wang, Gang Chen, Yan Lu | UI-Evol: Automatic Knowledge Evolving for Computer Use Agents | null | null | null | null | cs.HC cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | External knowledge has played a crucial role in the recent development of computer use agents. We identify a critical knowledge-execution gap: retrieved knowledge often fails to translate into effective real-world task execution. Our analysis shows even 90\% correct knowledge yields only 41\% execution success rate. To bridge this gap, we propose UI-Evol, a plug-and-play module for autonomous GUI knowledge evolution. UI-Evol consists of two stages: a Retrace Stage that extracts faithful objective action sequences from actual agent-environment interactions, and a Critique Stage that refines existing knowledge by comparing these sequences against external references. We conduct comprehensive experiments on the OSWorld benchmark with the state-of-the-art Agent S2. Our results demonstrate that UI-Evol not only significantly boosts task performance but also addresses a previously overlooked issue of high behavioral standard deviation in computer use agents, leading to superior performance on computer use tasks and substantially improved agent reliability. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:32:05 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhang",
"Ziyun",
""
],
[
"Liu",
"Xinyi",
""
],
[
"Zhang",
"Xiaoyi",
""
],
[
"Wang",
"Jun",
""
],
[
"Chen",
"Gang",
""
],
[
"Lu",
"Yan",
""
]
] |
2505.21965 | Guo-Zhao Liao | Guo-Zhao Liao, Xiao-Feng Gong, Wei Liu and Hing Cheung So | Target Localization with Coprime Multistatic MIMO Radar via Coupled Canonical Polyadic Decomposition Based on Joint Eigenvalue Decomposition | null | null | null | null | eess.SP | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper investigates target localization using a multistatic multiple-input multiple-output (MIMO) radar system with two distinct coprime array configurations: coprime L-shaped arrays and coprime planar arrays. The observed signals are modeled as tensors that admit a coupled canonical polyadic decomposition (C-CPD) model. For each configuration, a C-CPD method is presented based on joint eigenvalue decomposition (J-EVD). This computational framework includes (semi-)algebraic and optimization-based C-CPD algorithms and target localization that fuses direction-of-arrivals (DOAs) information to calculate the optimal position of each target. Specifically, the proposed (semi-)algebraic methods exploit the rotational invariance of the Vandermonde structure in coprime arrays, similar to the multiple invariance property of \added{estimation of signal parameters via rotational invariance techniques} (ESPRIT), which transforms the model into a J-EVD problem and reduces computational complexity. The study also investigates the working conditions of the algorithm to understand model identifiability. Additionally, the proposed method does not rely on prior knowledge of non-orthogonal probing waveforms and is effective in challenging underdetermined scenarios. Experimental results demonstrate that our method outperforms existing tensor-based approaches in both accuracy and computational efficiency. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:34:13 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Liao",
"Guo-Zhao",
""
],
[
"Gong",
"Xiao-Feng",
""
],
[
"Liu",
"Wei",
""
],
[
"So",
"Hing Cheung",
""
]
] |
2505.21966 | Aditya Gunturu | Aditya Gunturu, Ben Pearman, Keiichi Ihara, Morteza Faraji, Bryan Wang, Rubaiat Habib Kazi, and Ryo Suzuki | MapStory: LLM-Powered Text-Driven Map Animation Prototyping with Human-in-the-Loop Editing | 16 pages and 15 figures | null | null | null | cs.HC cs.AI cs.CL cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce MapStory, an LLM-powered animation authoring tool that generates editable map animation sequences directly from natural language text. Given a user-written script, MapStory leverages an agentic architecture to automatically produce a scene breakdown, which decomposes the script into key animation building blocks such as camera movements, visual highlights, and animated elements. Our system includes a researcher component that accurately queries geospatial information by leveraging an LLM with web search, enabling the automatic extraction of relevant regions, paths, and coordinates while allowing users to edit and query for changes or additional information to refine the results. Additionally, users can fine-tune parameters of these blocks through an interactive timeline editor. We detail the system's design and architecture, informed by formative interviews with professional animators and an analysis of 200 existing map animation videos. Our evaluation, which includes expert interviews (N=5) and a usability study (N=12), demonstrates that MapStory enables users to create map animations with ease, facilitates faster iteration, encourages creative exploration, and lowers barriers to creating map-centric stories. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:36:08 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Gunturu",
"Aditya",
""
],
[
"Pearman",
"Ben",
""
],
[
"Ihara",
"Keiichi",
""
],
[
"Faraji",
"Morteza",
""
],
[
"Wang",
"Bryan",
""
],
[
"Kazi",
"Rubaiat Habib",
""
],
[
"Suzuki",
"Ryo",
""
]
] |
2505.21967 | Usman Naseem | Juan Ren, Mark Dras, Usman Naseem | Seeing the Threat: Vulnerabilities in Vision-Language Models to Adversarial Attack | Preprint | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Large Vision-Language Models (LVLMs) have shown remarkable capabilities across a wide range of multimodal tasks. However, their integration of visual inputs introduces expanded attack surfaces, thereby exposing them to novel security vulnerabilities. In this work, we conduct a systematic representational analysis to uncover why conventional adversarial attacks can circumvent the safety mechanisms embedded in LVLMs. We further propose a novel two stage evaluation framework for adversarial attacks on LVLMs. The first stage differentiates among instruction non compliance, outright refusal, and successful adversarial exploitation. The second stage quantifies the degree to which the model's output fulfills the harmful intent of the adversarial prompt, while categorizing refusal behavior into direct refusals, soft refusals, and partial refusals that remain inadvertently helpful. Finally, we introduce a normative schema that defines idealized model behavior when confronted with harmful prompts, offering a principled target for safety alignment in multimodal systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:43:39 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ren",
"Juan",
""
],
[
"Dras",
"Mark",
""
],
[
"Naseem",
"Usman",
""
]
] |
2505.21968 | Hyejeong Ryu | Hyejeong Ryu | Enhanced SIRRT*: A Structure-Aware RRT* for 2D Path Planning with Hybrid Smoothing and Bidirectional Rewiring | null | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Sampling-based motion planners such as Rapidly-exploring Random Tree* (RRT*) and its informed variant IRRT* are widely used for optimal path planning in complex environments. However, these methods often suffer from slow convergence and high variance due to their reliance on random sampling, particularly when initial solution discovery is delayed. This paper presents Enhanced SIRRT* (E-SIRRT*), a structure-aware planner that improves upon the original SIRRT* framework by introducing two key enhancements: hybrid path smoothing and bidirectional rewiring. Hybrid path smoothing refines the initial path through spline fitting and collision-aware correction, while bidirectional rewiring locally optimizes tree connectivity around the smoothed path to improve cost propagation. Experimental results demonstrate that E-SIRRT* consistently outperforms IRRT* and SIRRT* in terms of initial path quality, convergence rate, and robustness across 100 trials. Unlike IRRT*, which exhibits high variability due to stochastic initialization, E-SIRRT* achieves repeatable and efficient performance through deterministic skeleton-based initialization and structural refinement. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:45:25 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ryu",
"Hyejeong",
""
]
] |
2505.21970 | Bikai Gao | Bikai Gao | Constraints on the strength of first-order phase transition in the low density region | null | null | null | null | nucl-th | http://creativecommons.org/licenses/by/4.0/ | We systematically studied the first-order phase transitions in the low-density region of neutron star matter using a unified framework combining the parity doublet model for hadronic matter, an NJL-type quark model for high-density quark matter, and a model-independent intermediate region treatment. By confronting with recent neutron star observation constraints, we find it possible to have the early phase transitions (PT) at low densities ($n_B \leq 2n_0$) in a way consistent with existing neutron star observations. We also establish the upper limits of the PT strength for each chiral invariant mass in parity doublet model. We find that larger chiral invariant mass lead to smaller allowed phase transition strengths, providing new insights and clue into nucleon mass origin. This connection between the microscopic structure of nucleons and macroscopic neutron star properties shows how astrophysical observations can constrain fundamental QCD mechanisms underlying hadron mass generation. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:48:19 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Gao",
"Bikai",
""
]
] |
2505.21971 | Chan-Byoung Chae | Robert W. Heath, Jr., Joseph Carlson, Nitish Vikas Deshpande, Miguel Rodrigo Castellanos, Mohamed Akrout, Chan-Byoung Chae | The Tri-Hybrid MIMO Architecture | 7 pages, 6 figures | null | null | null | cs.IT cs.NI math.IT | http://creativecommons.org/licenses/by/4.0/ | We present an evolution of multiple-input multiple-output (MIMO) wireless communications known as the tri-hybrid MIMO architecture. In this framework, the traditional operations of linear precoding at the transmitter are distributed across digital beamforming, analog beamforming, and reconfigurable antennas. Compared with the hybrid MIMO architecture, which combines digital and analog beamforming, the tri-hybrid approach introduces a third layer of electromagnetic beamforming through antenna reconfigurability. This added layer offers a pathway to scale MIMO spatial dimensions, important for 6G systems operating in centimeter-wave bands, where the tension between larger bandwidths and infrastructure reuse necessitates ultra-large antenna arrays. We introduce the key features of the tri-hybrid architecture by (i)~reviewing the benefits and challenges of communicating with reconfigurable antennas, (ii)~examining tradeoffs between spectral and energy efficiency enabled by reconfigurability, and (iii)~exploring configuration challenges across the three layers. Overall, the tri-hybrid MIMO architecture offers a new approach for integrating emerging antenna technologies in the MIMO precoding framework. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:50:03 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Heath,",
"Robert W.",
"Jr."
],
[
"Carlson",
"Joseph",
""
],
[
"Deshpande",
"Nitish Vikas",
""
],
[
"Castellanos",
"Miguel Rodrigo",
""
],
[
"Akrout",
"Mohamed",
""
],
[
"Chae",
"Chan-Byoung",
""
]
] |
2505.21972 | Patrick Vossler | Patrick Vossler, Fan Xia, Yifan Mai, Jean Feng | Judging LLMs on a Simplex | 28 pages, 7 figures | null | null | null | cs.LG cs.AI stat.ML | http://creativecommons.org/licenses/by/4.0/ | Automated evaluation of free-form outputs from large language models (LLMs) is challenging because many distinct answers can be equally valid. A common practice is to use LLMs themselves as judges, but the theoretical properties of this approach are not yet well understood. We show that a geometric framework that represents both judges and candidates as points on a probability simplex can provide helpful insight on what is or is not identifiable using LLM judges. Our theoretical analysis uncovers a "phase transition" in ranking identifiability: for binary scoring systems, true rankings are identifiable even with weak judges under mild assumptions, while rankings become non-identifiable for three or more scoring levels even with infinite data, absent additional prior knowledge. This non-identifiability highlights how uncertainty in rankings stems from not only aleatoric uncertainty (i.e., inherent stochasticity in the data) but also epistemic uncertainty regarding which assumptions hold, an aspect that has received limited attention until now. To integrate both types of uncertainty, we use Bayesian inference to encode assumptions as priors and conduct sensitivity analysis of ranking estimates and credible intervals. Empirical evaluations across multiple benchmarks demonstrate that Bayesian inference yields more accurate rankings and substantially improves coverage rates. These results underscore the importance of taking a more holistic approach to uncertainty quantification when using LLMs as judges. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:50:41 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Vossler",
"Patrick",
""
],
[
"Xia",
"Fan",
""
],
[
"Mai",
"Yifan",
""
],
[
"Feng",
"Jean",
""
]
] |
2505.21973 | Linyu Li | Linyu Li, Zhi Jin, Yichi Zhang, Dongming Jin, Chengfeng Dou, Yuanpeng He, Xuan Zhang, Haiyan Zhao | Towards Structure-aware Model for Multi-modal Knowledge Graph Completion | null | null | null | null | cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Knowledge graphs (KGs) play a key role in promoting various multimedia and AI applications. However, with the explosive growth of multi-modal information, traditional knowledge graph completion (KGC) models cannot be directly applied. This has attracted a large number of researchers to study multi-modal knowledge graph completion (MMKGC). Since MMKG extends KG to the visual and textual domains, MMKGC faces two main challenges: (1) how to deal with the fine-grained modality information interaction and awareness; (2) how to ensure the dominant role of graph structure in multi-modal knowledge fusion and deal with the noise generated by other modalities during modality fusion. To address these challenges, this paper proposes a novel MMKGC model named TSAM, which integrates fine-grained modality interaction and dominant graph structure to form a high-performance MMKGC framework. Specifically, to solve the challenges, TSAM proposes the Fine-grained Modality Awareness Fusion method (FgMAF), which uses pre-trained language models to better capture fine-grained semantic information interaction of different modalities and employs an attention mechanism to achieve fine-grained modality awareness and fusion. Additionally, TSAM presents the Structure-aware Contrastive Learning method (SaCL), which utilizes two contrastive learning approaches to align other modalities more closely with the structured modality. Extensive experiments show that the proposed TSAM model significantly outperforms existing MMKGC models on widely used multi-modal datasets. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 04:50:58 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Li",
"Linyu",
""
],
[
"Jin",
"Zhi",
""
],
[
"Zhang",
"Yichi",
""
],
[
"Jin",
"Dongming",
""
],
[
"Dou",
"Chengfeng",
""
],
[
"He",
"Yuanpeng",
""
],
[
"Zhang",
"Xuan",
""
],
[
"Zhao",
"Haiyan",
""
]
] |
2505.21975 | Weiguang Zhang | Weiguang Zhang, Huangcheng Lu, Maizhen Ning, Xiaowei Huang, Wei Wang, Kaizhu Huang, Qiufeng Wang | DvD: Unleashing a Generative Paradigm for Document Dewarping via Coordinates-based Diffusion Model | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Document dewarping aims to rectify deformations in photographic document images, thus improving text readability, which has attracted much attention and made great progress, but it is still challenging to preserve document structures. Given recent advances in diffusion models, it is natural for us to consider their potential applicability to document dewarping. However, it is far from straightforward to adopt diffusion models in document dewarping due to their unfaithful control on highly complex document images (e.g., 2000$\times$3000 resolution). In this paper, we propose DvD, the first generative model to tackle document \textbf{D}ewarping \textbf{v}ia a \textbf{D}iffusion framework. To be specific, DvD introduces a coordinate-level denoising instead of typical pixel-level denoising, generating a mapping for deformation rectification. In addition, we further propose a time-variant condition refinement mechanism to enhance the preservation of document structures. In experiments, we find that current document dewarping benchmarks can not evaluate dewarping models comprehensively. To this end, we present AnyPhotoDoc6300, a rigorously designed large-scale document dewarping benchmark comprising 6,300 real image pairs across three distinct domains, enabling fine-grained evaluation of dewarping models. Comprehensive experiments demonstrate that our proposed DvD can achieve state-of-the-art performance with acceptable computational efficiency on multiple metrics across various benchmarks including DocUNet, DIR300, and AnyPhotoDoc6300. The new benchmark and code will be publicly available. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 05:05:51 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhang",
"Weiguang",
""
],
[
"Lu",
"Huangcheng",
""
],
[
"Ning",
"Maizhen",
""
],
[
"Huang",
"Xiaowei",
""
],
[
"Wang",
"Wei",
""
],
[
"Huang",
"Kaizhu",
""
],
[
"Wang",
"Qiufeng",
""
]
] |
2505.21976 | Fuxin Guan | Fuxin Guan, Zemeng Lin, Sixin Chen, Xinhua Wen, and Shuang Zhang | High-order virtual gain for optical loss compensation in plasmonic metamaterials | null | null | null | null | physics.optics | http://creativecommons.org/licenses/by/4.0/ | Metamaterials exhibit extraordinary properties yet suffer from pronounced wave dissipation, particularly in optical imaging and sensing systems. Recent advances leveraging complex frequency wave excitations with virtual gain effect, synthesized by multi-monochromatic waves, offer promising solutions for optical loss compensation. However, this approach faces limitations in extreme loss scenarios. The complex frequency wave requires sufficient virtual gain, i.e., temporal attenuation, to offset material loss, inevitably triggering rapid signal decay to zero before reaching a quasi-static state. To address this challenge, we introduce synthetic waves of high-order virtual gain to slow down the decay rate while preserving the loss compensation efficiency. We experimentally demonstrate 20-fold noise suppression in plasmonic resonance systems compared to conventional complex frequency excitations. This approach exhibits broad applicability across diverse fields, including imaging, biosensing, and integrated photonic signal processing. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 05:07:34 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Guan",
"Fuxin",
""
],
[
"Lin",
"Zemeng",
""
],
[
"Chen",
"Sixin",
""
],
[
"Wen",
"Xinhua",
""
],
[
"Zhang",
"Shuang",
""
]
] |
2505.21977 | Khoa Ta | Khoa Ta | Homology of Rook-Brauer Algebras and Motzkin Algebras | null | null | null | null | math.RA math.RT | http://creativecommons.org/licenses/by/4.0/ | Using the technique of inductive resolution introduced in arXiv:2303.07979, we prove that the homology of Rook-Brauer Algebra, interpreted as appropriate Tor-group, is isomorphic to that of symmetric group for all degrees under the assumption that $\epsilon$ in $R$ is invertible; furthermore, we also prove the homology of the Motzkin algebras vanishes in positive degrees under the same assumption. These results thereby establish homological stability of both algebras. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 05:09:35 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ta",
"Khoa",
""
]
] |
2505.21978 | Zengyao Man | Wanfu Gao, Zengyao Man, Zebin He, Yuhao Tang, Jun Gao and Kunpeng Liu | Two-Stage Feature Generation with Transformer and Reinforcement Learning | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Feature generation is a critical step in machine learning, aiming to enhance model performance by capturing complex relationships within the data and generating meaningful new features. Traditional feature generation methods heavily rely on domain expertise and manual intervention, making the process labor-intensive and challenging to adapt to different scenarios. Although automated feature generation techniques address these issues to some extent, they often face challenges such as feature redundancy, inefficiency in feature space exploration, and limited adaptability to diverse datasets and tasks. To address these problems, we propose a Two-Stage Feature Generation (TSFG) framework, which integrates a Transformer-based encoder-decoder architecture with Proximal Policy Optimization (PPO). The encoder-decoder model in TSFG leverages the Transformer's self-attention mechanism to efficiently represent and transform features, capturing complex dependencies within the data. PPO further enhances TSFG by dynamically adjusting the feature generation strategy based on task-specific feedback, optimizing the process for improved performance and adaptability. TSFG dynamically generates high-quality feature sets, significantly improving the predictive performance of machine learning models. Experimental results demonstrate that TSFG outperforms existing state-of-the-art methods in terms of feature quality and adaptability. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 05:11:59 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Gao",
"Wanfu",
""
],
[
"Man",
"Zengyao",
""
],
[
"He",
"Zebin",
""
],
[
"Tang",
"Yuhao",
""
],
[
"Gao",
"Jun",
""
],
[
"Liu",
"Kunpeng",
""
]
] |
2505.21979 | Fakhraddin Alwajih | Fakhraddin Alwajih, Samar Mohamed Magdy, Abdellah El Mekki, Omer Nacar, Youssef Nafea, Safaa Taher Abdelfadil, Abdulfattah Mohammed Yahya, Hamzah Luqman, Nada Almarwani, Samah Aloufi, Baraah Qawasmeh, Houdaifa Atou, Serry Sibaee, Hamzah A. Alsayadi, Walid Al-Dhabyani, Maged S. Al-shaibani, Aya El aatar, Nour Qandos, Rahaf Alhamouri, Samar Ahmad, Razan Khassib, Lina Hamad, Mohammed Anwar AL-Ghrawi, Fatimah Alshamari, Cheikh Malainine, Doaa Qawasmeh, Aminetou Yacoub, Tfeil moilid, Ruwa AbuHweidi, Ahmed Aboeitta, Vatimetou Mohamed Lemin, Reem Abdel-Salam, Ahlam Bashiti, Adel Ammar, Aisha Alansari, Ahmed Ashraf, Nora Alturayeif, Sara Shatnawi, Alcides Alcoba Inciarte, AbdelRahim A. Elmadany, Mohamedou cheikh tourad, Ismail Berrada, Mustafa Jarrar, Shady Shehata, Muhammad Abdul-Mageed | Pearl: A Multimodal Culturally-Aware Arabic Instruction Dataset | https://github.com/UBC-NLP/pearl | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Mainstream large vision-language models (LVLMs) inherently encode cultural biases, highlighting the need for diverse multimodal datasets. To address this gap, we introduce Pearl, a large-scale Arabic multimodal dataset and benchmark explicitly designed for cultural understanding. Constructed through advanced agentic workflows and extensive human-in-the-loop annotations by 45 annotators from across the Arab world, Pearl comprises over K multimodal examples spanning ten culturally significant domains covering all Arab countries. We further provide two robust evaluation benchmarks Pearl and Pearl-Lite along with a specialized subset Pearl-X explicitly developed to assess nuanced cultural variations. Comprehensive evaluations on state-of-the-art open and proprietary LVLMs demonstrate that reasoning-centric instruction alignment substantially improves models' cultural grounding compared to conventional scaling methods. Pearl establishes a foundational resource for advancing culturally-informed multimodal modeling research. All datasets and benchmarks are publicly available. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 05:14:47 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Alwajih",
"Fakhraddin",
""
],
[
"Magdy",
"Samar Mohamed",
""
],
[
"Mekki",
"Abdellah El",
""
],
[
"Nacar",
"Omer",
""
],
[
"Nafea",
"Youssef",
""
],
[
"Abdelfadil",
"Safaa Taher",
""
],
[
"Yahya",
"Abdulfattah Mohammed",
""
],
[
"Luqman",
"Hamzah",
""
],
[
"Almarwani",
"Nada",
""
],
[
"Aloufi",
"Samah",
""
],
[
"Qawasmeh",
"Baraah",
""
],
[
"Atou",
"Houdaifa",
""
],
[
"Sibaee",
"Serry",
""
],
[
"Alsayadi",
"Hamzah A.",
""
],
[
"Al-Dhabyani",
"Walid",
""
],
[
"Al-shaibani",
"Maged S.",
""
],
[
"aatar",
"Aya El",
""
],
[
"Qandos",
"Nour",
""
],
[
"Alhamouri",
"Rahaf",
""
],
[
"Ahmad",
"Samar",
""
],
[
"Khassib",
"Razan",
""
],
[
"Hamad",
"Lina",
""
],
[
"AL-Ghrawi",
"Mohammed Anwar",
""
],
[
"Alshamari",
"Fatimah",
""
],
[
"Malainine",
"Cheikh",
""
],
[
"Qawasmeh",
"Doaa",
""
],
[
"Yacoub",
"Aminetou",
""
],
[
"moilid",
"Tfeil",
""
],
[
"AbuHweidi",
"Ruwa",
""
],
[
"Aboeitta",
"Ahmed",
""
],
[
"Lemin",
"Vatimetou Mohamed",
""
],
[
"Abdel-Salam",
"Reem",
""
],
[
"Bashiti",
"Ahlam",
""
],
[
"Ammar",
"Adel",
""
],
[
"Alansari",
"Aisha",
""
],
[
"Ashraf",
"Ahmed",
""
],
[
"Alturayeif",
"Nora",
""
],
[
"Shatnawi",
"Sara",
""
],
[
"Inciarte",
"Alcides Alcoba",
""
],
[
"Elmadany",
"AbdelRahim A.",
""
],
[
"tourad",
"Mohamedou cheikh",
""
],
[
"Berrada",
"Ismail",
""
],
[
"Jarrar",
"Mustafa",
""
],
[
"Shehata",
"Shady",
""
],
[
"Abdul-Mageed",
"Muhammad",
""
]
] |
2505.21980 | Hilay Shah | Hilay Shah, Freeke van de Voort, Amit Seta, Christoph Federrath | Understanding gas mixing in the circumgalactic medium | 23 pages, 16 figures, submitted to MNRAS, under review | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study gas mixing in a simulated Milky Way-mass galaxy's circumgalactic medium (CGM) using cosmological zoom-in simulations. We insert tracer dyes in the CGM with different gas flows (shearing, coherent, and static) and diverse physical properties to track gas mixing and find that shearing flows generally exhibit a higher dye spread. We then correlate the extent and shape of dye spread with local gas properties for a generalised understanding of gas mixing. The velocity dispersion and magnitude of traceless symmetric shear (shear deformation) tensors in small regions (<5 kpc) around the dye injection locations most accurately predict the extent of dye spread after 200 Myr (Spearman rank 0.87). We use this to determine diffusivity calibration constants for subgrid-scale mixing models. While the dye shape after 200 Myr aligns well with velocity and magnetic field dispersion, the best alignment occurs with the dispersion of eigenvectors representing stretching (from traceless symmetric shear tensor) and plane-of-rotation (from antisymmetric shear or vorticity tensor) in large regions (>10 kpc) around the dye injection locations. Therefore, shear-related statistics and velocity dispersion are the properties best able to predict both the extent and shape of mixed gas in our simulations. Our dyes spread roughly linearly with time, suggesting superdiffusive behaviour in the CGM, potentially due to turbulent and large-scale coherent flows. We calculate physical Reynolds numbers of 10^2-10^3 using gas properties in the CGM of our simulations, which a resolution of a few hundred pc could resolve, but our 1 kpc resolution leads to significant numerical mixing. However, our correlation results are robust to changes in resolution. These results can be used to predict diffusion coefficients from local gas properties to model processes like magnetic field diffusion, heat transport, and metal mixing. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 05:17:59 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Shah",
"Hilay",
""
],
[
"van de Voort",
"Freeke",
""
],
[
"Seta",
"Amit",
""
],
[
"Federrath",
"Christoph",
""
]
] |
2505.21981 | Weiyu Liu | Weiyu Liu, Neil Nie, Ruohan Zhang, Jiayuan Mao, Jiajun Wu | Learning Compositional Behaviors from Demonstration and Language | Presented at CoRL 2024 and as an Oral Presentation at the 2024 CoRL LEAP Workshop. The first two authors contributed equally. The last two authors jointly advised the project. For videos and additional results, visit: https://blade-bot.github.io/ | null | null | null | cs.RO cs.AI cs.CL cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce Behavior from Language and Demonstration (BLADE), a framework for long-horizon robotic manipulation by integrating imitation learning and model-based planning. BLADE leverages language-annotated demonstrations, extracts abstract action knowledge from large language models (LLMs), and constructs a library of structured, high-level action representations. These representations include preconditions and effects grounded in visual perception for each high-level action, along with corresponding controllers implemented as neural network-based policies. BLADE can recover such structured representations automatically, without manually labeled states or symbolic definitions. BLADE shows significant capabilities in generalizing to novel situations, including novel initial states, external state perturbations, and novel goals. We validate the effectiveness of our approach both in simulation and on real robots with a diverse set of objects with articulated parts, partial observability, and geometric constraints. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 05:19:59 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Liu",
"Weiyu",
""
],
[
"Nie",
"Neil",
""
],
[
"Zhang",
"Ruohan",
""
],
[
"Mao",
"Jiayuan",
""
],
[
"Wu",
"Jiajun",
""
]
] |
2505.21982 | Aaditya Shankar Majumder | Aaditya Shankar Majumder | Eye-Tracking and Biometric Feedback in UX Research: Measuring User Engagement and Cognitive Load | 4 pages | null | null | null | cs.HC | http://creativecommons.org/licenses/by/4.0/ | User experience research often uses surveys and interviews, which may miss subconscious user interactions. This study explores eye-tracking and biometric feedback as tools to assess user engagement and cognitive load in digital interfaces. These methods measure gaze behavior and bodily responses, providing an objective complement to qualitative insights. Using empirical evidence, practical applications, and advancements from 2023-2025, we present experimental data, describe our methodology, and place our work within foundational and recent literature. We address challenges like data interpretation, ethical issues, and technological integration. These tools are key for advancing UX design in complex digital environments. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 05:21:29 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Majumder",
"Aaditya Shankar",
""
]
] |
2505.21983 | Martha M. L\'opez-Guti\'errez | M. M. L\'opez-Guti\'errez, H. Bravo-Alfaro, P. T. Rahna, G. A. Mamon, Y. L. Jaff\'e, L. F. Madrigal-Ayala, E. Acosta-Espinoza | Tracing Galaxy Evolution in infalling galaxies of Abell 496: From Starburst to Quenching | 16 pages, 11 figures | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | During the fall of late-type galaxies into clusters, they can experiment a variety of evolutionary mechanisms according their local environment. Consequently, studying the UV emission and the cold gas of late-type galaxies provide key insights in the evolution of short-lived starburst and galaxy quenching. In this work, we conduted a study of two 28' fields observed with UVIT-AstroSat in the central region of the Abell cluster A496 ($z=0.033$), including HI, data from NRAO VLA. We reported 22 cluster members detected in FUV; all of them are detected in HI, or have upper limits for the HI-mass. We find our FUV detected galaxies generally have higher specific star formation rates than other star forming galaxies. Most of the FUV galaxies with masses above 10$^9 \mathrm{M}_{\odot}$,and showing high sSFR have no close neighbors, pointing at RPS as the dominant mechanism affecting them. In contrast, most of the low-mass FUV objects present at least one companion, suggesting that tidal interactions also play an important role in the triggering of infalling galaxies. Combining the FUV-SFR with the HI properties of the observed galaxies in A496 we identify an evolutionary sequence consisting of five stages: (1) Pre-triggering, (2) Initial SF-triggering, (3) Peak of star-formation, (4) SF-fading, and (5) SF-quenching. During this path, normal gas-rich objects reach a gas-deficiency phase with SFR well below the main sequence. This process, prior to becoming a full passive galaxy, can be accomplished within a few 10$^{8}$ yr. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 05:22:17 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"López-Gutiérrez",
"M. M.",
""
],
[
"Bravo-Alfaro",
"H.",
""
],
[
"Rahna",
"P. T.",
""
],
[
"Mamon",
"G. A.",
""
],
[
"Jaffé",
"Y. L.",
""
],
[
"Madrigal-Ayala",
"L. F.",
""
],
[
"Acosta-Espinoza",
"E.",
""
]
] |
2505.21984 | Satoshi Takada | Ken Okamura, Yosuke Sato, Satoshi Takada | Stress distribution in elastic disks with a hole under uniaxial compression | 4 pages, 3 figures, Powders & Grains 2025 | null | null | null | cond-mat.soft physics.class-ph | http://creativecommons.org/licenses/by/4.0/ | This paper investigates the stress and displacement distribution in a two-dimensional elastic hollow disk subjected to distributed diametric loading, extending our previous analysis of concentrated loading [Okamura et al. Strength Mater. 57, 102-114 (2025)]. The study provides deeper insights into the mechanical behavior of materials such as concrete and rock by examining the effects of load distribution on stress localization and displacement patterns. Using elastodynamic theory, we derive the static stress distributions and identify key differences from the concentrated loading case, particularly in the locations and magnitudes of stress extrema. This work contributes to a more comprehensive understanding of stress behavior in elastic disks under realistic loading conditions. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 05:22:38 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Okamura",
"Ken",
""
],
[
"Sato",
"Yosuke",
""
],
[
"Takada",
"Satoshi",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.