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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2505.22507 | Marco Bee | Marco Bee and Flavio Santi | Modeling and estimating skewed and heavy-tailed populations via unsupervised mixture models | null | null | null | null | stat.ME | http://creativecommons.org/licenses/by/4.0/ | We develop an unsupervised mixture model for non-negative, skewed and heavy-tailed data, such as losses in actuarial and risk management applications. The mixture has a lognormal component, which is usually appropriate for the body of the distribution, and a Pareto-type tail, aimed at accommodating the largest observations, since the lognormal tail often decays too fast. We show that maximum likelihood estimation can be performed by means of the EM algorithm and that the model is quite flexible in fitting data from different data-generating processes. Simulation experiments and a real-data application to automobiles claims suggest that the approach is equivalent in terms of goodness-of-fit, but easier to estimate, with respect to two existing distributions with similar features. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 15:54:40 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bee",
"Marco",
""
],
[
"Santi",
"Flavio",
""
]
] |
2505.22508 | George Ntavazlis Katsaros | George N. Katsaros and Konstantinos Nikitopoulos | Toward Fully Neuromorphic Receivers for Ultra-Power Efficient Communications | Accepted IEEE SPAWC 2025 | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | Neuromorphic computing, inspired by biological neural systems, has emerged as a promising approach for ultra-energy-efficient data processing by leveraging analog neuron structures and spike-based computation. However, its application in communication systems remains largely unexplored, with existing efforts mainly focused on mapping isolated communication algorithms onto spiking networks, often accompanied by substantial, traditional computational overhead due to transformations required to adapt problems to the spiking paradigm. In this work, we take a fundamentally different route and, for the first time, propose a fully neuromorphic communication receiver by applying neuromorphic principles directly in the analog domain from the very start of the receiver processing chain. Specifically, we examine a simple transmission scenario: a BPSK receiver with repetition coding, and show that we can achieve joint detection and decoding entirely through spiking signals. Our approach demonstrates error-rate performance gains over conventional digital realizations with power consumption on the order of microwatts, comparable with a single very low-resolution Analog-to-Digital Converter (ADC) utilized in digital receivers. To maintain performance under varying noise conditions, we also introduce a novel noise-tracking mechanism that dynamically adjusts neural parameters during transmission. Finally, we discuss the key challenges and directions toward ultra-efficient neuromorphic transceivers. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 15:57:16 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Katsaros",
"George N.",
""
],
[
"Nikitopoulos",
"Konstantinos",
""
]
] |
2505.22509 | Zhonglin Xie | Zhonglin Xie, Yiman Fong, Haoran Yuan, Zaiwen Wen | Accelerating Optimization via Differentiable Stopping Time | null | null | null | null | cs.LG math.OC | http://creativecommons.org/licenses/by/4.0/ | Optimization is an important module of modern machine learning applications. Tremendous efforts have been made to accelerate optimization algorithms. A common formulation is achieving a lower loss at a given time. This enables a differentiable framework with respect to the algorithm hyperparameters. In contrast, its dual, minimizing the time to reach a target loss, is believed to be non-differentiable, as the time is not differentiable. As a result, it usually serves as a conceptual framework or is optimized using zeroth-order methods. To address this limitation, we propose a differentiable stopping time and theoretically justify it based on differential equations. An efficient algorithm is designed to backpropagate through it. As a result, the proposed differentiable stopping time enables a new differentiable formulation for accelerating algorithms. We further discuss its applications, such as online hyperparameter tuning and learning to optimize. Our proposed methods show superior performance in comprehensive experiments across various problems, which confirms their effectiveness. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 15:59:13 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Xie",
"Zhonglin",
""
],
[
"Fong",
"Yiman",
""
],
[
"Yuan",
"Haoran",
""
],
[
"Wen",
"Zaiwen",
""
]
] |
2505.22510 | Parbati Senapati | Parbati Senapati, Ajay Kumar and Prakash Parida | Enhanced thermopower in two-dimensional ruthenium dichalcogenides $RuX_2$ (X = S, Se): a first-principles study | null | null | 10.1088/1402-4896/add29e | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Transition metal dichalcogenides (TMDs) have garnered attention for their potential in thermoelectric applications due to their unique electronic properties and tunable bandgaps. In this study, we systematically explore the electronic and thermoelectric properties of $T^{\prime}-RuX_2$ (X = S, Se) using first-principles calculations and semi-classical Boltzmann transport equations. Our findings confirm that $T^{\prime}-RuX_2$ is energetically and mechanically stable, with high thermopower values such that $T^{\prime}-RuS_2$ exhibits a Seebeck coefficient of $2685~\mu V/K$ for hole doping and $2585~\mu V/K$ for electron doping, while $T^{\prime}-RuSe_2$ shows values of $1515~\mu V/K$ and $1533~\mu V/K$ for hole and electron doping, respectively. Both materials exhibit reasonable power factors and $ZT$ values, with p-type $T^{\prime}-RuS_2$ and $T^{\prime}-RuSe_2$ achieving maximum ZT values of 0.85 and 0.87, respectively, at 1200~K along the y-direction. These results highlight $T^{\prime}$-$RuS_2$ and $T^{\prime}$-$RuSe_2$ as promising candidates for high-temperature TMD-based thermoelectric devices. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:00:12 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Senapati",
"Parbati",
""
],
[
"Kumar",
"Ajay",
""
],
[
"Parida",
"Prakash",
""
]
] |
2505.22512 | Saniya Khan | Saniya Khan, Richard I. Anderson, Sylvia Ekstr\"om, Cyril Georgy, Louise Breuval | The stellar evolution perspective on the metallicity dependence of classical Cepheid Leavitt laws | 20 pages, 11 figures, submitted to A&A. Comments are welcome! | null | null | null | astro-ph.SR astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The impact of metallicity on the Cepheid Leavitt law (LL) and, in turn, the Hubble constant, has been the subject of much recent debate. Here, we present a comprehensive analysis of metallicity effects on Cepheid LLs based on synthetic Cepheid populations computed using Geneva models and the SYCLIST tool. We computed 296 co-eval populations in the age range of 5-300 Myr for metallicities representative of the Sun, the LMC, and the SMC ($Z \in [0.014, 0.006, 0.002]$). We computed LLs in fourteen optical-to-infrared passbands and five reddening-free Wesenheit magnitudes. All Cepheid populations take into account distributions of rotation rates and companion stars. We show excellent agreement between the predicted populations and key observational constraints from the literature. Our simulations predict a significant LL slope-metallicity dependence ($\beta_{\rm M} > 0$) that renders LLs steeper at lower metallicity at all wavelengths. Importantly, $\beta_{\rm M} \ne 0$ implies that the intercept-metallicity dependence, $\alpha_{\rm M}$, depends on pivot period; an issue not previously considered. Comparison with $\alpha_{\rm M}$ measurements in individual passbands reported in the literature yields acceptable agreement on the order of agreement found among different observational studies. The wavelength dependence and magnitude of the disagreement suggests a possible origin in reddening-related systematics. Conversely, we report excellent agreement between our $\alpha_{\rm M} = -0.20 \pm 0.03$ mag dex$^{-1}$ and the value determined by the SH0ES distance ladder in the reddening-free H-band Wesenheit magnitude ($-0.217 \pm 0.046$), the currently tightest and conceptually simplest empirical constraint. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:01:45 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Khan",
"Saniya",
""
],
[
"Anderson",
"Richard I.",
""
],
[
"Ekström",
"Sylvia",
""
],
[
"Georgy",
"Cyril",
""
],
[
"Breuval",
"Louise",
""
]
] |
2505.22513 | Nicholas Teh | Bradley Phillips, Edith Elkind, Nicholas Teh, Tomasz W\k{a}s | Strengthening Proportionality in Temporal Voting | null | null | null | null | cs.GT cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study proportional representation in the framework of temporal voting with approval ballots. Prior work adapted basic proportional representation concepts -- justified representation (JR), proportional JR (PJR), and extended JR (EJR) -- from the multiwinner setting to the temporal setting. Our work introduces and examines ways of going beyond EJR. Specifically, we consider stronger variants of JR, PJR, and EJR, and introduce temporal adaptations of more demanding multiwinner axioms, such as EJR+, full JR (FJR), full proportional JR (FPJR), and the Core. For each of these concepts, we investigate its existence and study its relationship to existing notions, thereby establishing a rich hierarchy of proportionality concepts. Notably, we show that two of our proposed axioms -- EJR+ and FJR -- strengthen EJR while remaining satisfiable in every temporal election. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:02:52 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Phillips",
"Bradley",
""
],
[
"Elkind",
"Edith",
""
],
[
"Teh",
"Nicholas",
""
],
[
"Wąs",
"Tomasz",
""
]
] |
2505.22514 | Jakub Paw{\l}owski | J. Pawlowski, P. Tarasiuk, J. Tuziemski, L. Pawela, and B. Gardas | Closing the Quantum-Classical Scaling Gap in Approximate Optimization | null | null | null | null | quant-ph physics.comp-ph | http://creativecommons.org/licenses/by/4.0/ | In a recent study (Ref. [1]), quantum annealing was reported to exhibit a scaling advantage for approximately solving Quadratic Unconstrained Binary Optimization (QUBO). However, this claim critically depends on the choice of classical reference algorithm -- Parallel Tempering with Isoenergetic Cluster Moves (PT-ICM). Here, we reassess these findings with different classical paradigm -- Simulated Bifurcation Machine (SBM) -- that harnesses nonlinear Hamiltonian dynamics. By leveraging chaotic behavior rather than thermal fluctuations, SBM achieves comparable or superior scaling performance, effectively closing the previously reported quantum-classical gap. We show that small problem sizes analyzed in [1] are insufficient for inferring asymptotic scaling, due to sensitivity to runtime and hardware-specific factors. By extending the benchmark to larger instances -- beyond current quantum annealing capabilities -- we establish strong classical scaling behavior. And as a result, we conclude that it is unlikely that current generation of quantum annealers, can demonstrate supremacy in discrete approximate optimization under operationally meaningful conditions. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:02:57 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Pawlowski",
"J.",
""
],
[
"Tarasiuk",
"P.",
""
],
[
"Tuziemski",
"J.",
""
],
[
"Pawela",
"L.",
""
],
[
"Gardas",
"B.",
""
]
] |
2505.22515 | Haoran Wang | Haoran Wang, Guanyu Chen, Bohan Li, Hankun Wang, Yiwei Guo, Zhihan Li, Xie Chen, Kai Yu | Towards General Discrete Speech Codec for Complex Acoustic Environments: A Study of Reconstruction and Downstream Task Consistency | Initial Upload | null | null | null | cs.SD eess.AS | http://creativecommons.org/licenses/by/4.0/ | Neural speech codecs excel in reconstructing clean speech signals; however, their efficacy in complex acoustic environments and downstream signal processing tasks remains underexplored. In this study, we introduce a novel benchmark named Environment-Resilient Speech Codec Benchmark (ERSB) to systematically evaluate whether neural speech codecs are environment-resilient. Specifically, we assess two key capabilities: (1) robust reconstruction, which measures the preservation of both speech and non-speech acoustic details, and (2) downstream task consistency, which ensures minimal deviation in downstream signal processing tasks when using reconstructed speech instead of the original. Our comprehensive experiments reveal that complex acoustic environments significantly degrade signal reconstruction and downstream task consistency. This work highlights the limitations of current speech codecs and raises a future direction that improves them for greater environmental resilience. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:03:03 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wang",
"Haoran",
""
],
[
"Chen",
"Guanyu",
""
],
[
"Li",
"Bohan",
""
],
[
"Wang",
"Hankun",
""
],
[
"Guo",
"Yiwei",
""
],
[
"Li",
"Zhihan",
""
],
[
"Chen",
"Xie",
""
],
[
"Yu",
"Kai",
""
]
] |
2505.22516 | Martina Toscani | Martina Toscani, Luca Broggi, Alberto Sesana and Elena Maria Rossi | Updated predictions for gravitational wave emission from TDEs for next generation observatories | 10 figures, 10 pages, submitted to A&A | null | null | null | astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we investigate the gravitational wave (GW) emission from stars tidally disrupted by black holes (TDEs), using a semi-analytical approach. Contrary to previous works where this signal is modeled as a monochromatic burst, we here take into account all its harmonic components. On top of this, we also extend the analysis to a population of repeated-partial TDEs, where the star undergoes multiple passages around the black hole before complete disruption. For both populations, we estimate the rate of individual GW-detections considering future observatories like LISA and a potential deci-Hertz (dHz) mission, and derive the GW background from these sources. Our conclusions, even if more conservative, are consistent with previous results presented in literature. In fact, full disruptions of stars will not be seen by LISA but will be important targets for dHz observatories. In contrast, GWs from repeated disruptions will not be detectable in the near future. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:03:37 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Toscani",
"Martina",
""
],
[
"Broggi",
"Luca",
""
],
[
"Sesana",
"Alberto",
""
],
[
"Rossi",
"Elena Maria",
""
]
] |
2505.22517 | Yimeng Gu | Yimeng Gu, Zhao Tong, Ignacio Castro, Shu Wu, Gareth Tyson | Multi-MLLM Knowledge Distillation for Out-of-Context News Detection | null | null | null | null | cs.CL cs.MM | http://creativecommons.org/licenses/by/4.0/ | Multimodal out-of-context news is a type of misinformation in which the image is used outside of its original context. Many existing works have leveraged multimodal large language models (MLLMs) for detecting out-of-context news. However, observing the limited zero-shot performance of smaller MLLMs, they generally require label-rich fine-tuning and/or expensive API calls to GPT models to improve the performance, which is impractical in low-resource scenarios. In contrast, we aim to improve the performance of small MLLMs in a more label-efficient and cost-effective manner. To this end, we first prompt multiple teacher MLLMs to generate both label predictions and corresponding rationales, which collectively serve as the teachers' knowledge. We then introduce a two-stage knowledge distillation framework to transfer this knowledge to a student MLLM. In Stage 1, we apply LoRA fine-tuning to the student model using all training data. In Stage 2, we further fine-tune the student model using both LoRA fine-tuning and DPO on the data points where teachers' predictions conflict. This two-stage strategy reduces annotation costs and helps the student model uncover subtle patterns in more challenging cases. Experimental results demonstrate that our approach achieves state-of-the-art performance using less than 10% labeled data. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:03:41 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Gu",
"Yimeng",
""
],
[
"Tong",
"Zhao",
""
],
[
"Castro",
"Ignacio",
""
],
[
"Wu",
"Shu",
""
],
[
"Tyson",
"Gareth",
""
]
] |
2505.22518 | Agnideep Aich | Agnideep Aich, Ashit Baran Aich and Bruce Wade | IGNIS: A Neural Network Framework for Robust Parameter Estimation in Archimedean Copulas | Under review | null | null | null | stat.ML cs.LG | http://creativecommons.org/licenses/by/4.0/ | Parameter estimation for Archimedean copulas remains a challenging problem, particularly for the recently developed A1 and A2 families that exhibit complex dependency structures. Traditional methods, such as the Method of Moments (MoM), Maximum Likelihood Estimation (MLE), and Maximum Pseudo-Likelihood (MPL), often struggle due to issues of non-monotonic relationship with dependency measures such as Kendall's tau (as in the case of A1) and numerical instability. In this paper, we present the IGNIS Network, a novel, unified neural framework that learns a direct mapping from observable dependency measures to copula parameters, thereby overcoming the limitations of classical approaches. Our approach is trained on simulated data spanning five Archimedean copula families including Clayton, Gumbel, Frank, A1, and A2, ensuring its general applicability across the entire family. Extensive simulation studies demonstrate that the IGNIS Network reduces estimation errors compared to MoM, while inherently enforcing parameter constraints through theory-guided post-processing. We further validate the practical utility of our method on diverse real-world datasets, including financial returns (AAPL-MSFT), healthcare metrics (CDC Diabetes indicators), and environmental measurements (PM2.5 air quality). Our results underscore the transformative potential of neural methods for robust and accurate dependence modeling in modern applications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:04:17 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Aich",
"Agnideep",
""
],
[
"Aich",
"Ashit Baran",
""
],
[
"Wade",
"Bruce",
""
]
] |
2505.22519 | Mateusz Wasilewski | Kristin Courtney, Priyanga Ganesan, Mateusz Wasilewski | Connectivity for quantum graphs via quantum adjacency operators | 16 pages | null | null | null | math.OA math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | Connectivity is a fundamental property of quantum graphs, previously studied in the operator system model for matrix quantum graphs and via graph homomorphisms in the quantum adjacency matrix model. In this paper, we develop an algebraic characterization of connectivity for general quantum graphs within the quantum adjacency matrix framework. Our approach extends earlier results to the non-tracial setting and beyond regular quantum graphs. We utilize a quantum Perron-Frobenius theorem that provides a spectral characterization of connectivity, and we further characterize connectivity in terms of the irreducibility of the quantum adjacency matrix and the nullity of the associated graph Laplacian. These results are obtained using the KMS inner product, which unifies and generalizes existing formulations. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:05:35 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Courtney",
"Kristin",
""
],
[
"Ganesan",
"Priyanga",
""
],
[
"Wasilewski",
"Mateusz",
""
]
] |
2505.22520 | Nimmi Das Anthuparambil | Nimmi Das Anthuparambil, Michelle Dargasz, Sonja Timmermann, Anita Girelli, Sebastian Retzbach, Johannes M\"oller, Wonhyuk Jo, Agha Mohammad Raza, Aliaksandr Leonau, James Wrigley, Frederik Unger, Maddalena Bin, Prince Prabhu Rajaiah, Iason Andronis, William Ch\`evremont, J\"org Hallmann, Angel Rodriguez-Fernandez, Jan-Etienne Pudell, Felix Brausse, Ulrike Boesenberg, Mohamed Youssef, Roman Shayduk, Rustam Rysov, Anders Madsen, Felix Lehmk\"uhler, Michael Paulus, Fajun Zhang, Fivos Perakis, Frank Schreiber, and Christian Gutt | Softness and Hydrodynamic Interactions Regulate Lipoprotein Transport in Crowded Yolk Environments | 59 pages | null | null | null | cond-mat.soft | http://creativecommons.org/licenses/by/4.0/ | Low-density lipoproteins (LDLs) serve as nutrient reservoirs in egg yolk for embryonic development and as promising drug carriers. Both roles critically depend on their mobility in densely crowded biological environments. Under these crowded conditions, diffusion is hindered by transient confinement within dynamic cages formed by neighboring particles, driven by solvent-mediated hydrodynamic interactions and memory effects -- phenomena that have remained challenging to characterize computationally and experimentally. Here, we employ megahertz X-ray photon correlation spectroscopy to directly probe the cage dynamics of LDLs in yolk-plasma across various concentrations. We find that LDLs undergo anomalous diffusion, experiencing $\approx$ 100-fold reduction in self-diffusion at high concentrations compared to dilute solutions. This drastic slowing-down is attributed to a combination of hydrodynamic interactions, direct particle-particle interactions, and the inherent softness of LDL particles. Despite reduced dynamics, yolk-plasma remains as a liquid, yet sluggish, balancing dense packing, structural stability, and fluidity essential for controlled lipid release during embryogenesis. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:07:49 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Anthuparambil",
"Nimmi Das",
""
],
[
"Dargasz",
"Michelle",
""
],
[
"Timmermann",
"Sonja",
""
],
[
"Girelli",
"Anita",
""
],
[
"Retzbach",
"Sebastian",
""
],
[
"Möller",
"Johannes",
""
],
[
"Jo",
"Wonhyuk",
""
],
[
"Raza",
"Agha Mohammad",
""
],
[
"Leonau",
"Aliaksandr",
""
],
[
"Wrigley",
"James",
""
],
[
"Unger",
"Frederik",
""
],
[
"Bin",
"Maddalena",
""
],
[
"Rajaiah",
"Prince Prabhu",
""
],
[
"Andronis",
"Iason",
""
],
[
"Chèvremont",
"William",
""
],
[
"Hallmann",
"Jörg",
""
],
[
"Rodriguez-Fernandez",
"Angel",
""
],
[
"Pudell",
"Jan-Etienne",
""
],
[
"Brausse",
"Felix",
""
],
[
"Boesenberg",
"Ulrike",
""
],
[
"Youssef",
"Mohamed",
""
],
[
"Shayduk",
"Roman",
""
],
[
"Rysov",
"Rustam",
""
],
[
"Madsen",
"Anders",
""
],
[
"Lehmkühler",
"Felix",
""
],
[
"Paulus",
"Michael",
""
],
[
"Zhang",
"Fajun",
""
],
[
"Perakis",
"Fivos",
""
],
[
"Schreiber",
"Frank",
""
],
[
"Gutt",
"Christian",
""
]
] |
2505.22521 | Yunbo Liu | Chao Wang, Chuanhao Nie, Yunbo Liu | Evaluating Supervised Learning Models for Fraud Detection: A Comparative Study of Classical and Deep Architectures on Imbalanced Transaction Data | 5 pages. Chao Wang, Chuanhao Nie, and Yunbo Liu contributed equally to this work. Corresponding author: Yunbo Liu ([email protected]). Submitted to the 3rd International Conference on Management Innovation and Economy Development (MIED 2025), Chongqing, China | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Fraud detection remains a critical task in high-stakes domains such as finance and e-commerce, where undetected fraudulent transactions can lead to significant economic losses. In this study, we systematically compare the performance of four supervised learning models - Logistic Regression, Random Forest, Light Gradient Boosting Machine (LightGBM), and a Gated Recurrent Unit (GRU) network - on a large-scale, highly imbalanced online transaction dataset. While ensemble methods such as Random Forest and LightGBM demonstrated superior performance in both overall and class-specific metrics, Logistic Regression offered a reliable and interpretable baseline. The GRU model showed strong recall for the minority fraud class, though at the cost of precision, highlighting a trade-off relevant for real-world deployment. Our evaluation emphasizes not only weighted averages but also per-class precision, recall, and F1-scores, providing a nuanced view of each model's effectiveness in detecting rare but consequential fraudulent activity. The findings underscore the importance of choosing models based on the specific risk tolerance and operational needs of fraud detection systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:08:04 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wang",
"Chao",
""
],
[
"Nie",
"Chuanhao",
""
],
[
"Liu",
"Yunbo",
""
]
] |
2505.22522 | Yuan Zhang | Yuan Zhang and Feng Chen and Yaolei Qi and Guanyu Yang and Huazhu Fu | PathFL: Multi-Alignment Federated Learning for Pathology Image Segmentation | 17 pages, 5 figures; Accepted by MedIA | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Pathology image segmentation across multiple centers encounters significant challenges due to diverse sources of heterogeneity including imaging modalities, organs, and scanning equipment, whose variability brings representation bias and impedes the development of generalizable segmentation models. In this paper, we propose PathFL, a novel multi-alignment Federated Learning framework for pathology image segmentation that addresses these challenges through three-level alignment strategies of image, feature, and model aggregation. Firstly, at the image level, a collaborative style enhancement module aligns and diversifies local data by facilitating style information exchange across clients. Secondly, at the feature level, an adaptive feature alignment module ensures implicit alignment in the representation space by infusing local features with global insights, promoting consistency across heterogeneous client features learning. Finally, at the model aggregation level, a stratified similarity aggregation strategy hierarchically aligns and aggregates models on the server, using layer-specific similarity to account for client discrepancies and enhance global generalization. Comprehensive evaluations on four sets of heterogeneous pathology image datasets, encompassing cross-source, cross-modality, cross-organ, and cross-scanner variations, validate the effectiveness of our PathFL in achieving better performance and robustness against data heterogeneity. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:09:02 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhang",
"Yuan",
""
],
[
"Chen",
"Feng",
""
],
[
"Qi",
"Yaolei",
""
],
[
"Yang",
"Guanyu",
""
],
[
"Fu",
"Huazhu",
""
]
] |
2505.22523 | Yuhui Yuan | Junwen Chen and Heyang Jiang and Yanbin Wang and Keming Wu and Ji Li and Chao Zhang and Keiji Yanai and Dong Chen and Yuhui Yuan | PrismLayers: Open Data for High-Quality Multi-Layer Transparent Image Generative Models | Homepage: https://prism-layers.github.io/ | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Generating high-quality, multi-layer transparent images from text prompts can unlock a new level of creative control, allowing users to edit each layer as effortlessly as editing text outputs from LLMs. However, the development of multi-layer generative models lags behind that of conventional text-to-image models due to the absence of a large, high-quality corpus of multi-layer transparent data. In this paper, we address this fundamental challenge by: (i) releasing the first open, ultra-high-fidelity PrismLayers (PrismLayersPro) dataset of 200K (20K) multilayer transparent images with accurate alpha mattes, (ii) introducing a trainingfree synthesis pipeline that generates such data on demand using off-the-shelf diffusion models, and (iii) delivering a strong, open-source multi-layer generation model, ART+, which matches the aesthetics of modern text-to-image generation models. The key technical contributions include: LayerFLUX, which excels at generating high-quality single transparent layers with accurate alpha mattes, and MultiLayerFLUX, which composes multiple LayerFLUX outputs into complete images, guided by human-annotated semantic layout. To ensure higher quality, we apply a rigorous filtering stage to remove artifacts and semantic mismatches, followed by human selection. Fine-tuning the state-of-the-art ART model on our synthetic PrismLayersPro yields ART+, which outperforms the original ART in 60% of head-to-head user study comparisons and even matches the visual quality of images generated by the FLUX.1-[dev] model. We anticipate that our work will establish a solid dataset foundation for the multi-layer transparent image generation task, enabling research and applications that require precise, editable, and visually compelling layered imagery. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:09:33 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Chen",
"Junwen",
""
],
[
"Jiang",
"Heyang",
""
],
[
"Wang",
"Yanbin",
""
],
[
"Wu",
"Keming",
""
],
[
"Li",
"Ji",
""
],
[
"Zhang",
"Chao",
""
],
[
"Yanai",
"Keiji",
""
],
[
"Chen",
"Dong",
""
],
[
"Yuan",
"Yuhui",
""
]
] |
2505.22524 | Zijing Ou | Chinmay Pani, Zijing Ou, Yingzhen Li | Test-Time Alignment of Discrete Diffusion Models with Sequential Monte Carlo | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Discrete diffusion models have become highly effective across various domains. However, real-world applications often require the generative process to adhere to certain constraints but without task-specific fine-tuning. To this end, we propose a training-free method based on Sequential Monte Carlo (SMC) to sample from the reward-aligned target distribution at the test time. Our approach leverages twisted SMC with an approximate locally optimal proposal, obtained via a first-order Taylor expansion of the reward function. To address the challenge of ill-defined gradients in discrete spaces, we incorporate a Gumbel-Softmax relaxation, enabling efficient gradient-based approximation within the discrete generative framework. Empirical results on both synthetic datasets and image modelling validate the effectiveness of our approach. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:12:03 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Pani",
"Chinmay",
""
],
[
"Ou",
"Zijing",
""
],
[
"Li",
"Yingzhen",
""
]
] |
2505.22525 | Steffi Chern | Ethan Chern, Zhulin Hu, Steffi Chern, Siqi Kou, Jiadi Su, Yan Ma, Zhijie Deng, Pengfei Liu | Thinking with Generated Images | null | null | null | null | cs.CV cs.AI cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | We present Thinking with Generated Images, a novel paradigm that fundamentally transforms how large multimodal models (LMMs) engage with visual reasoning by enabling them to natively think across text and vision modalities through spontaneous generation of intermediate visual thinking steps. Current visual reasoning with LMMs is constrained to either processing fixed user-provided images or reasoning solely through text-based chain-of-thought (CoT). Thinking with Generated Images unlocks a new dimension of cognitive capability where models can actively construct intermediate visual thoughts, critique their own visual hypotheses, and refine them as integral components of their reasoning process. We demonstrate the effectiveness of our approach through two complementary mechanisms: (1) vision generation with intermediate visual subgoals, where models decompose complex visual tasks into manageable components that are generated and integrated progressively, and (2) vision generation with self-critique, where models generate an initial visual hypothesis, analyze its shortcomings through textual reasoning, and produce refined outputs based on their own critiques. Our experiments on vision generation benchmarks show substantial improvements over baseline approaches, with our models achieving up to 50% (from 38% to 57%) relative improvement in handling complex multi-object scenarios. From biochemists exploring novel protein structures, and architects iterating on spatial designs, to forensic analysts reconstructing crime scenes, and basketball players envisioning strategic plays, our approach enables AI models to engage in the kind of visual imagination and iterative refinement that characterizes human creative, analytical, and strategic thinking. We release our open-source suite at https://github.com/GAIR-NLP/thinking-with-generated-images. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:12:45 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Chern",
"Ethan",
""
],
[
"Hu",
"Zhulin",
""
],
[
"Chern",
"Steffi",
""
],
[
"Kou",
"Siqi",
""
],
[
"Su",
"Jiadi",
""
],
[
"Ma",
"Yan",
""
],
[
"Deng",
"Zhijie",
""
],
[
"Liu",
"Pengfei",
""
]
] |
2505.22526 | Fei Qin | Fei Qin, Zhanxin Hao, Jifan Yu, Zhiyuan Liu, Yu Zhang | AI instructional agent improves student's perceived learner control and learning outcome: empirical evidence from a randomized controlled trial | null | null | null | null | cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study examines the impact of an AI instructional agent on students' perceived learner control and academic performance in a medium demanding course with lecturing as the main teaching strategy. Based on a randomized controlled trial, three instructional conditions were compared: a traditional human teacher, a self-paced MOOC with chatbot support, and an AI instructional agent capable of delivering lectures and responding to questions in real time. Students in the AI instructional agent group reported significantly higher levels of perceived learner control compared to the other groups. They also completed the learning task more efficiently and engaged in more frequent interactions with the instructional system. Regression analyzes showed that perceived learner control positively predicted post-test performance, with behavioral indicators such as reduced learning time and higher interaction frequency supporting this relationship. These findings suggest that AI instructional agents, when designed to support personalized pace and responsive interaction, can enhance both students' learning experience and learning outcomes. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:13:27 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Qin",
"Fei",
""
],
[
"Hao",
"Zhanxin",
""
],
[
"Yu",
"Jifan",
""
],
[
"Liu",
"Zhiyuan",
""
],
[
"Zhang",
"Yu",
""
]
] |
2505.22527 | Agnideep Aich | Agnideep Aich, Ashit Aich and Bruce Wade | Symplectic Generative Networks (SGNs): A Hamiltonian Framework for Invertible Deep Generative Modeling | Submitted | null | null | null | stat.ML cs.LG | http://creativecommons.org/licenses/by/4.0/ | We introduce the Symplectic Generative Network (SGN), a deep generative model that leverages Hamiltonian mechanics to construct an invertible, volume-preserving mapping between a latent space and the data space. By endowing the latent space with a symplectic structure and modeling data generation as the time evolution of a Hamiltonian system, SGN achieves exact likelihood evaluation without incurring the computational overhead of Jacobian determinant calculations. In this work, we provide a rigorous mathematical foundation for SGNs through a comprehensive theoretical framework that includes: (i) complete proofs of invertibility and volume preservation, (ii) a formal complexity analysis with theoretical comparisons to Variational Autoencoders and Normalizing Flows, (iii) strengthened universal approximation results with quantitative error bounds, (iv) an information-theoretic analysis based on the geometry of statistical manifolds, and (v) an extensive stability analysis with adaptive integration guarantees. These contributions highlight the fundamental advantages of SGNs and establish a solid foundation for future empirical investigations and applications to complex, high-dimensional data. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:13:36 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Aich",
"Agnideep",
""
],
[
"Aich",
"Ashit",
""
],
[
"Wade",
"Bruce",
""
]
] |
2505.22528 | Elena Denisova | Elena Denisova | $\delta$-invariants of log Fano planes | 70 pp | null | null | null | math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We compute the $\delta$-invariant for pairs $(\mathbb{P}^2, \lambda C_d)$, where $C_d$ is a plane curve of degree $d \leq 4$. These computations provide new examples of $K$-stable and $K$-semistable log Fano pairs, and contribute to the study of $K$-stability of log Fano varieties via the Abban-Zhuang method, which reduces higher-dimensional problems to the surface case. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:14:36 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Denisova",
"Elena",
""
]
] |
2505.22529 | Niclas Bernhoff | Niclas Bernhoff | Discrete Boltzmann Equation for Anyons | 13 pages; MAT-DYN-NET 24 | null | null | null | math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A semi-classical approach to the study of the evolution of anyonic excitations--elementary particles with fractional statistics, complementing bosons and fermions--is through the Boltzmann equation for anyons. This work reviews a discretized version--a system of partial differential equations--of such a quantum equation. Trend to equilibrium is studied for a planar stationary system, as well as the spatially homogeneous system. Essential properties of the linearized operator are proven, implying that results for general steady half-space problems for the discrete Boltzmann equation in a slab geometry can be applied. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:15:31 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bernhoff",
"Niclas",
""
]
] |
2505.22530 | Andrei Vesnin | Siqi Ding, Fengchun Lei, Wei Lin, Andrei Vesnin | Amalgamations along surfaces with boundary in a handlebody | 18 pages, 11 figures | null | null | null | math.GT | http://creativecommons.org/licenses/by/4.0/ | Let M be a connected orientable 3-manifold, and F a compact connected orientable surface properly embedded in M. If F cuts M into two connected 3-manifolds X and Y, that is, M=X \cup_F Y, we say that M is an amalgamation of X and Y along F; and if F cuts M into a connected 3-manifold X, we say that M is a self-amalgamation of X along F. A characterization of an amalgamation of two handlebodies along a surface, incompressible in both, to be a handlebody was obtained by Lei, Liu, Li, and Vesnin. The case of amalgamation of two handelbodies along a compressional surface was studdied by Xu, Fang, and Lei. In the present paper, a characterization of an amalgamation and self-amalgamation of a handlebody to be a handlebody is given. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:16:58 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ding",
"Siqi",
""
],
[
"Lei",
"Fengchun",
""
],
[
"Lin",
"Wei",
""
],
[
"Vesnin",
"Andrei",
""
]
] |
2505.22531 | Andr\'es Molina-Markham | Andres Molina-Markham, Luis Robaina, Sean Steinle, Akash Trivedi, Derek Tsui, Nicholas Potteiger, Lauren Brandt, Ransom Winder, Ahmed Ridley | Training RL Agents for Multi-Objective Network Defense Tasks | null | null | null | null | cs.LG cs.AI cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Open-ended learning (OEL) -- which emphasizes training agents that achieve broad capability over narrow competency -- is emerging as a paradigm to develop artificial intelligence (AI) agents to achieve robustness and generalization. However, despite promising results that demonstrate the benefits of OEL, applying OEL to develop autonomous agents for real-world cybersecurity applications remains a challenge.
We propose a training approach, inspired by OEL, to develop autonomous network defenders. Our results demonstrate that like in other domains, OEL principles can translate into more robust and generalizable agents for cyber defense. To apply OEL to network defense, it is necessary to address several technical challenges. Most importantly, it is critical to provide a task representation approach over a broad universe of tasks that maintains a consistent interface over goals, rewards and action spaces. This way, the learning agent can train with varying network conditions, attacker behaviors, and defender goals while being able to build on previously gained knowledge.
With our tools and results, we aim to fundamentally impact research that applies AI to solve cybersecurity problems. Specifically, as researchers develop gyms and benchmarks for cyber defense, it is paramount that they consider diverse tasks with consistent representations, such as those we propose in our work. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:18:21 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Molina-Markham",
"Andres",
""
],
[
"Robaina",
"Luis",
""
],
[
"Steinle",
"Sean",
""
],
[
"Trivedi",
"Akash",
""
],
[
"Tsui",
"Derek",
""
],
[
"Potteiger",
"Nicholas",
""
],
[
"Brandt",
"Lauren",
""
],
[
"Winder",
"Ransom",
""
],
[
"Ridley",
"Ahmed",
""
]
] |
2505.22532 | Robert Altmann | R. Altmann and B. D\"orich and C. Zimmer | Gautschi-type and implicit-explicit integrators for constrained wave equations | null | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper deals with the construction and analysis of two integrators for (semi-linear) second-order partial differential-algebraic equations of semi-explicit type. More precisely, we consider an implicit-explicit Crank-Nicolson scheme as well as an exponential integrator of Gautschi type. For this, well-known wave integrators for unconstrained systems are combined with techniques known from the field of differential-algebraic equations. The result are efficient time stepping schemes, which are provable of second order. Moreover, we discuss the practical implementation of the Gautschi-type method, which involves the solution of certain saddle point problems. The theoretical results are verified by numerical experiments for the the wave equation with kinetic boundary conditions. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:19:22 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Altmann",
"R.",
""
],
[
"Dörich",
"B.",
""
],
[
"Zimmer",
"C.",
""
]
] |
2505.22533 | Pallavi Bhardwaj | Pallavi Bhardwaj, Caitlin Jones, Lasse Dierich, Aleksandar Vu\v{c}kovi\'c | TabularQGAN: A Quantum Generative Model for Tabular Data | 18 pages,8 figures and 4 tables | null | null | null | cs.LG cs.AI quant-ph | http://creativecommons.org/licenses/by-nc-sa/4.0/ | In this paper, we introduce a novel quantum generative model for synthesizing tabular data. Synthetic data is valuable in scenarios where real-world data is scarce or private, it can be used to augment or replace existing datasets. Real-world enterprise data is predominantly tabular and heterogeneous, often comprising a mixture of categorical and numerical features, making it highly relevant across various industries such as healthcare, finance, and software. We propose a quantum generative adversarial network architecture with flexible data encoding and a novel quantum circuit ansatz to effectively model tabular data. The proposed approach is tested on the MIMIC III healthcare and Adult Census datasets, with extensive benchmarking against leading classical models, CTGAN, and CopulaGAN. Experimental results demonstrate that our quantum model outperforms classical models by an average of 8.5% with respect to an overall similarity score from SDMetrics, while using only 0.072% of the parameters of the classical models. Additionally, we evaluate the generalization capabilities of the models using two custom-designed metrics that demonstrate the ability of the proposed quantum model to generate useful and novel samples. To our knowledge, this is one of the first demonstrations of a successful quantum generative model for handling tabular data, indicating that this task could be well-suited to quantum computers. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:19:39 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bhardwaj",
"Pallavi",
""
],
[
"Jones",
"Caitlin",
""
],
[
"Dierich",
"Lasse",
""
],
[
"Vučković",
"Aleksandar",
""
]
] |
2505.22534 | Ameek Malhotra | Aya Ghaleb, Ameek Malhotra, Gianmassimo Tasinato and Ivonne Zavala | Bayesian reconstruction of primordial perturbations from induced gravitational waves | 32 pages, 14 figures | null | null | null | astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | The formation of primordial black holes or other dark matter relics from amplified density fluctuations in the early universe may also generate scalar-induced gravitational waves (GW), carrying vital information about the primordial power spectrum and the early expansion history of our universe. We present a Bayesian approach aimed at reconstructing both the shape of the scalar power spectrum and the universe's equation of state from GW observations, using interpolating splines to flexibly capture features in the GW data. The optimal number of spline nodes is chosen via Bayesian evidence, aiming at balancing complexity of the model and the fidelity of the reconstruction. We test our method using both representative mock data and recent Pulsar Timing Array measurements, demonstrating that it can accurately reconstruct the curvature power spectrum as well as the underlying equation of state, if different from radiation. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:21:00 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ghaleb",
"Aya",
""
],
[
"Malhotra",
"Ameek",
""
],
[
"Tasinato",
"Gianmassimo",
""
],
[
"Zavala",
"Ivonne",
""
]
] |
2505.22536 | Neda Boroumand | Neda Boroumand, Adam Thorpe, Graeme Bart, Lu Wang, David N. Purschke, Giulio Vampa, Thomas Brabec | Quantum engineering of high harmonic generation | null | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In quantum sideband high harmonic generation (QSHHG), high harmonic generation is perturbed by a bright quantum field resulting in harmonic sidebands, with the intent to transfer non-classical properties from the quantum perturbation to the harmonic sidebands. So far, non-classical features have not been found in QSHHG yet. The closed form theory of QSHHG in atoms and solids developed here answers the question under which conditions non-classical features can be realized. QSHHG results in a multi-mode entanglement between harmonic sideband modes and perturbative quantum mode. A projective measurement on either creates a variety of non-classical states commonly used in quantum information science. This opens a pathway towards quantum engineering high harmonic generation as a short wavelength source for quantum information science. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:22:01 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Boroumand",
"Neda",
""
],
[
"Thorpe",
"Adam",
""
],
[
"Bart",
"Graeme",
""
],
[
"Wang",
"Lu",
""
],
[
"Purschke",
"David N.",
""
],
[
"Vampa",
"Giulio",
""
],
[
"Brabec",
"Thomas",
""
]
] |
2505.22537 | Maxence Wynen | Maxence Wynen, Pedro M. Gordaliza, Maxime Istasse, Anna St\"olting, Pietro Maggi, Beno\^it Macq, Meritxell Bach Cuadra | ConfLUNet: Multiple sclerosis lesion instance segmentation in presence of confluent lesions | null | null | null | null | eess.IV | http://creativecommons.org/licenses/by-sa/4.0/ | Accurate lesion-level segmentation on MRI is critical for multiple sclerosis (MS) diagnosis, prognosis, and disease monitoring. However, current evaluation practices largely rely on semantic segmentation post-processed with connected components (CC), which cannot separate confluent lesions (aggregates of confluent lesion units, CLUs) due to reliance on spatial connectivity. To address this misalignment with clinical needs, we introduce formal definitions of CLUs and associated CLU-aware detection metrics, and include them in an exhaustive instance segmentation evaluation framework. Within this framework, we systematically evaluate CC and post-processing-based Automated Confluent Splitting (ACLS), the only existing methods for lesion instance segmentation in MS. Our analysis reveals that CC consistently underestimates CLU counts, while ACLS tends to oversplit lesions, leading to overestimated lesion counts and reduced precision. To overcome these limitations, we propose ConfLUNet, the first end-to-end instance segmentation framework for MS lesions. ConfLUNet jointly optimizes lesion detection and delineation from a single FLAIR image. Trained on 50 patients, ConfLUNet significantly outperforms CC and ACLS on the held-out test set (n=13) in instance segmentation (Panoptic Quality: 42.0% vs. 37.5%/36.8%; p = 0.017/0.005) and lesion detection (F1: 67.3% vs. 61.6%/59.9%; p = 0.028/0.013). For CLU detection, ConfLUNet achieves the highest F1[CLU] (81.5%), improving recall over CC (+12.5%, p = 0.015) and precision over ACLS (+31.2%, p = 0.003). By combining rigorous definitions, new CLU-aware metrics, a reproducible evaluation framework, and the first dedicated end-to-end model, this work lays the foundation for lesion instance segmentation in MS. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:22:27 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wynen",
"Maxence",
""
],
[
"Gordaliza",
"Pedro M.",
""
],
[
"Istasse",
"Maxime",
""
],
[
"Stölting",
"Anna",
""
],
[
"Maggi",
"Pietro",
""
],
[
"Macq",
"Benoît",
""
],
[
"Cuadra",
"Meritxell Bach",
""
]
] |
2505.22538 | Paul Hofman | Paul Hofman, Yusuf Sale, Eyke H\"ullermeier | Uncertainty Quantification with Proper Scoring Rules: Adjusting Measures to Prediction Tasks | null | null | null | null | cs.LG stat.ML | http://creativecommons.org/licenses/by/4.0/ | We address the problem of uncertainty quantification and propose measures of total, aleatoric, and epistemic uncertainty based on a known decomposition of (strictly) proper scoring rules, a specific type of loss function, into a divergence and an entropy component. This leads to a flexible framework for uncertainty quantification that can be instantiated with different losses (scoring rules), which makes it possible to tailor uncertainty quantification to the use case at hand. We show that this flexibility is indeed advantageous. In particular, we analyze the task of selective prediction and show that the scoring rule should ideally match the task loss. In addition, we perform experiments on two other common tasks. For out-of-distribution detection, our results confirm that a widely used measure of epistemic uncertainty, mutual information, performs best. Moreover, in the setting of active learning, our measure of epistemic uncertainty based on the zero-one-loss consistently outperforms other uncertainty measures. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:22:53 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hofman",
"Paul",
""
],
[
"Sale",
"Yusuf",
""
],
[
"Hüllermeier",
"Eyke",
""
]
] |
2505.22539 | Petar Lukovic | Tim Engelbracht, Petar Lukovic, Tjark Behrens, Kai Lascheit, Ren\'e Zurbr\"ugg, Marc Pollefeys, Hermann Blum and Zuria Bauer | Spot-On: A Mixed Reality Interface for Multi-Robot Cooperation | null | null | null | null | cs.HC cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent progress in mixed reality (MR) and robotics is enabling increasingly sophisticated forms of human-robot collaboration. Building on these developments, we introduce a novel MR framework that allows multiple quadruped robots to operate in semantically diverse environments via a MR interface. Our system supports collaborative tasks involving drawers, swing doors, and higher-level infrastructure such as light switches. A comprehensive user study verifies both the design and usability of our app, with participants giving a "good" or "very good" rating in almost all cases. Overall, our approach provides an effective and intuitive framework for MR-based multi-robot collaboration in complex, real-world scenarios. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:23:28 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Engelbracht",
"Tim",
""
],
[
"Lukovic",
"Petar",
""
],
[
"Behrens",
"Tjark",
""
],
[
"Lascheit",
"Kai",
""
],
[
"Zurbrügg",
"René",
""
],
[
"Pollefeys",
"Marc",
""
],
[
"Blum",
"Hermann",
""
],
[
"Bauer",
"Zuria",
""
]
] |
2505.22540 | Tobias Brixner | Andreas M\"uller, Stefan Mueller, Tobias Brixner, Sebastian von Mammen | A Graph-Based Laser Path Solver Algorithm for Virtual Reality Laboratory Simulations | 35 pages, 8 figures, 2 tables, 1 algorithm | null | null | null | physics.optics physics.ed-ph | http://creativecommons.org/licenses/by/4.0/ | femtoPro is an interactive virtual reality (VR) laser laboratory balancing the contrasting challenges of accuracy and computational efficiency in optics simulations. It can simulate linear and nonlinear optical phenomena in real time, a task that pushes the boundaries of current consumer hardware. This paper details the concept, implementation, and evaluation of a dynamic graph-based solution tailored to the specific requirements and challenges of the simulation. Resource usage is optimized through a selective updating strategy that identifies and preserves laser paths unchanged between simulation frames, eliminating the need for unnecessary recalculations. Benchmarking of real-world scenarios confirms that our approach delivers a smooth user experience, even on mobile VR platforms with limited computing power. The methodologies, solutions and insights outlined in this paper may be applicable to other interactive, dynamic graph-based real-time simulations. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:23:32 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Müller",
"Andreas",
""
],
[
"Mueller",
"Stefan",
""
],
[
"Brixner",
"Tobias",
""
],
[
"von Mammen",
"Sebastian",
""
]
] |
2505.22541 | Vinitra Swamy | Vinitra Swamy | A Human-Centric Approach to Explainable AI for Personalized Education | PhD Thesis, EPFL (Computer Science) | null | null | null | cs.LG cs.CY | http://creativecommons.org/licenses/by/4.0/ | Deep neural networks form the backbone of artificial intelligence research, with potential to transform the human experience in areas ranging from autonomous driving to personal assistants, healthcare to education. However, their integration into the daily routines of real-world classrooms remains limited. It is not yet common for a teacher to assign students individualized homework targeting their specific weaknesses, provide students with instant feedback, or simulate student responses to a new exam question. While these models excel in predictive performance, this lack of adoption can be attributed to a significant weakness: the lack of explainability of model decisions, leading to a lack of trust from students, parents, and teachers. This thesis aims to bring human needs to the forefront of eXplainable AI (XAI) research, grounded in the concrete use case of personalized learning and teaching. We frame the contributions along two verticals: technical advances in XAI and their aligned human studies. We investigate explainability in AI for education, revealing systematic disagreements between post-hoc explainers and identifying a need for inherently interpretable model architectures. We propose four novel technical contributions in interpretability with a multimodal modular architecture (MultiModN), an interpretable mixture-of-experts model (InterpretCC), adversarial training for explainer stability, and a theory-driven LLM-XAI framework to present explanations to students (iLLuMinaTE), which we evaluate in diverse settings with professors, teachers, learning scientists, and university students. By combining empirical evaluations of existing explainers with novel architectural designs and human studies, our work lays a foundation for human-centric AI systems that balance state-of-the-art performance with built-in transparency and trust. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:23:48 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Swamy",
"Vinitra",
""
]
] |
2505.22542 | Hosein Gholami | Hosein Gholami, Lennart Kurth, Ugo Mire, Michael Buballa, Bernd-Jochen Schaefer | Renormalizing the Quark-Meson-Diquark Model | 26 pages, 5 figures, 2 Tables. Comments are welcomed | null | null | null | hep-ph hep-th nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a comprehensive study of the two-flavor Quark-Meson-Diquark (QMD) model by comparing a renormalization approach with a renormalization-group (RG) consistent mean-field formulation based on the functional renormalization group (FRG). The renormalized QMD allows analytical investigations of key quantities such as the zero-temperature diquark gap and the critical temperature for color superconductivity, ultimately reproducing the exact BCS relation in the high-density limit. We carry out the same analysis for different schemes of RG-consistent QMD models. We show that the RG-consistent approach yields a phase diagram and thermodynamic properties qualitatively similar to those of the renormalization model, provided both are embedded within a unified scheme that ensures consistent vacuum properties. In particular, both treatments recover the expected high-density features characterized by the BCS relation and the Stefan-Boltzmann limit. Our results highlight the relevance of renormalization and RG-consistent methods for accurately capturing the thermodynamics of QMD and related effective models with diquark degrees of freedom. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:24:49 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Gholami",
"Hosein",
""
],
[
"Kurth",
"Lennart",
""
],
[
"Mire",
"Ugo",
""
],
[
"Buballa",
"Michael",
""
],
[
"Schaefer",
"Bernd-Jochen",
""
]
] |
2505.22543 | Ziheng Jia | Ziheng Jia, Zicheng Zhang, Zeyu Zhang, Yingji Liang, Xiaorong Zhu, Chunyi Li, Jinliang Han, Haoning Wu, Bin Wang, Haoran Zhang, Guanyu Zhu, Qiyong Zhao, Xiaohong Liu, Guangtao Zhai, Xiongkuo Min | Scaling-up Perceptual Video Quality Assessment | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | The data scaling law has been shown to significantly enhance the performance of large multi-modal models (LMMs) across various downstream tasks. However, in the domain of perceptual video quality assessment (VQA), the potential of scaling law remains unprecedented due to the scarcity of labeled resources and the insufficient scale of datasets. To address this, we propose \textbf{OmniVQA}, an efficient framework designed to efficiently build high-quality, human-in-the-loop VQA multi-modal instruction databases (MIDBs). We then scale up to create \textbf{OmniVQA-Chat-400K}, the largest MIDB in the VQA field concurrently. Our focus is on the technical and aesthetic quality dimensions, with abundant in-context instruction data to provide fine-grained VQA knowledge. Additionally, we have built the \textbf{OmniVQA-MOS-20K} dataset to enhance the model's quantitative quality rating capabilities. We then introduce a \textbf{complementary} training strategy that effectively leverages the knowledge from datasets for quality understanding and quality rating tasks. Furthermore, we propose the \textbf{OmniVQA-FG (fine-grain)-Benchmark} to evaluate the fine-grained performance of the models. Our results demonstrate that our models achieve state-of-the-art performance in both quality understanding and rating tasks. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:24:52 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Jia",
"Ziheng",
""
],
[
"Zhang",
"Zicheng",
""
],
[
"Zhang",
"Zeyu",
""
],
[
"Liang",
"Yingji",
""
],
[
"Zhu",
"Xiaorong",
""
],
[
"Li",
"Chunyi",
""
],
[
"Han",
"Jinliang",
""
],
[
"Wu",
"Haoning",
""
],
[
"Wang",
"Bin",
""
],
[
"Zhang",
"Haoran",
""
],
[
"Zhu",
"Guanyu",
""
],
[
"Zhao",
"Qiyong",
""
],
[
"Liu",
"Xiaohong",
""
],
[
"Zhai",
"Guangtao",
""
],
[
"Min",
"Xiongkuo",
""
]
] |
2505.22544 | Kan Zhao | Kan Zhao, Hao Deng, Hua Chen, Nvsen Ma, Noah Oefele, Jiesen Guo, Xueling Cui, Chen Tang, Matthias J. Gutmann, Thomas Mueller, Yixi Su, Vladimir Hutanu, Changqing Jin and Philipp Gegenwart | Nonlinear time-reversal symmetry breaking in kagome spin ice HoAgGe | This manuscript contains 19 pages and 5 figures, with Supplemental Materials not included | null | null | null | cond-mat.str-el cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Kagome spin ice is an intriguing class of spin systems constituted by in-plane Ising spins with ferromagnetic interaction residing on the kagome lattice, theoretically predicted to host a plethora of magnetic transitions and excitations. In particular, different variants of kagome spin ice models can exhibit different sequences of symmetry breaking upon cooling from the paramagnetic to the fully ordered ground state. Recently, it has been demonstrated that the frustrated intermetallic HoAgGe stands as a faithful solid-state realization of kagome spin ice. Here we use single crystal neutron diffuse scattering to map the spin ordering of HoAgGe at various temperatures more accurately and surprisingly find that the ordering sequence appears to be different from previously known scenarios: From the paramagnetic state, the system first enters a partially ordered state with fluctuating magnetic charges, in contrast to a charge-ordered paramagnetic phase before reaching the fully ordered state. Through state-of-the-art Monte Carlo simulations and scaling analyses using a quasi-2D model for the distorted Kagome spin ice in HoAgGe, we elucidate a single three-dimensional (3D) XY phase transition into the ground state with broken time-reversal symmetry (TRS). However, the 3D XY transition has a long crossover tail before the fluctuating magnetic charges fully order. More interestingly, we find both experimentally and theoretically that the TRS breaking phase of HoAgGe features an unusual, hysteretic response: In spite of their vanishing magnetization, the two time-reversal partners are distinguished and selected by a nonlinear magnetic susceptibility tied to the kagome ice rule. Our discovery not only unveils a new symmetry breaking hierarchy of kagome spin ice, but also demonstrates the potential of TRS-breaking frustrated spin systems for information technology applications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:25:28 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhao",
"Kan",
""
],
[
"Deng",
"Hao",
""
],
[
"Chen",
"Hua",
""
],
[
"Ma",
"Nvsen",
""
],
[
"Oefele",
"Noah",
""
],
[
"Guo",
"Jiesen",
""
],
[
"Cui",
"Xueling",
""
],
[
"Tang",
"Chen",
""
],
[
"Gutmann",
"Matthias J.",
""
],
[
"Mueller",
"Thomas",
""
],
[
"Su",
"Yixi",
""
],
[
"Hutanu",
"Vladimir",
""
],
[
"Jin",
"Changqing",
""
],
[
"Gegenwart",
"Philipp",
""
]
] |
2505.22545 | Christine Bradley | Christine L. Bradley, Rami W. Wehbe, Matthew Smith, Sharmila Padmanabhan, Valerie Scott, David R. Thompson, Daniel W. Wilson, Pantazis Mouroulis, Robert O. Green, Christian Frankenberg | The Optical Design of the Carbon Investigation(Carbon-I) Imaging Spectrometer | 18 pages, 14 figures, IEEE Aerospace Conference 2025 | null | null | null | physics.optics astro-ph.IM | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The proposed Carbon Investigation (Carbon-I) Imaging Spectrometer is designed to measure variations of greenhouse gases in Earth's atmosphere. The instrument will survey the Earth from its own spacecraft at an altitude of approximately 610 km. It will use a coarse ground sampling distance (GSD) of <400 m in global mode for land and coastal monitoring and finer 35 m GSD in target mode to sample key regions. The identification and quantification of greenhouse gases require continuous spectral sampling over the 2040-2380 nm wavelength range with <1 nm spectral sampling. The proposed design builds upon Jet Propulsion Laboratory's (JPL) experience of spaceflight Dyson imaging spectrometers to achieve spectral sampling of 0.7 nm per pixel. This paper presents the proposed Carbon-I optical design comprised of a freeform three-mirror anastigmat telescope that couples to a F/2.2, highly uniform Dyson-inspired imaging spectrometer. The high uniformity and throughput enables Carbon-I to measure Earth's greenhouse gas concentrations with unprecedented precision and spatial sampling. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:25:29 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bradley",
"Christine L.",
""
],
[
"Wehbe",
"Rami W.",
""
],
[
"Smith",
"Matthew",
""
],
[
"Padmanabhan",
"Sharmila",
""
],
[
"Scott",
"Valerie",
""
],
[
"Thompson",
"David R.",
""
],
[
"Wilson",
"Daniel W.",
""
],
[
"Mouroulis",
"Pantazis",
""
],
[
"Green",
"Robert O.",
""
],
[
"Frankenberg",
"Christian",
""
]
] |
2505.22546 | Timothy Strobel | Wencheng Lu, Michael J. Hutcheon, Mads F. Hansen, Kapildeb Dolui, Shubham Sinha, Mihir R. Sahoo, Chris J. Pickard, Christoph Heil, Anna Pakhomova, Mohamed Mezouar, Dominik Daisenberger, Stella Chariton, Vitali Prakapenka, Matthew N. Julian, Rohit P. Prasankumar, Timothy A. Strobel | Prediction and Synthesis of Mg$_4$Pt$_3$H$_6$: A Metallic Complex Transition Metal Hydride Stabilized at Ambient Pressure | null | null | null | null | cond-mat.supr-con cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | The low-pressure stabilization of superconducting hydrides with high critical temperatures ($T_c$s) remains a significant challenge, and experimentally verified superconducting hydrides are generally constrained to a limited number of structural prototypes. Ternary transition-metal complex hydrides (hydrido complexes)-typically regarded as hydrogen storage materials-exhibit a large range of compounds stabilized at low pressure with recent predictions for high-$T_c$ superconductivity. Motivated by this class of materials, we investigated complex hydride formation in the Mg-Pt-H system, which has no known ternary hydride compounds. Guided by ab initio structural predictions, we successfully synthesized a novel complex transition-metal hydride, Mg$_4$Pt$_3$H$_6$, using laser-heated diamond anvil cells. The compound forms in a body-centered cubic structural prototype at moderate pressures between 8-25 GPa. Unlike the majority of known hydrido complexes, Mg$_4$Pt$_3$H$_6$ is metallic, with formal charge described as 4[Mg]$^{2+}$.3[PtH$_2$]$^{2-}$. X-ray diffraction (XRD) measurements obtained during decompression reveal that Mg$_4$Pt$_3$H$_6$ remains stable upon quenching to ambient conditions. Magnetic-field and temperature-dependent electrical transport measurements indicate ambient-pressure superconductivity with $T_c$ (50%) = 2.9 K, in reasonable agreement with theoretical calculations. These findings clarify the phase behavior in the Mg-Pt-H system and provide valuable insights for transition-metal complex hydrides as a new class of hydrogen-rich superconductors. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:28:16 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lu",
"Wencheng",
""
],
[
"Hutcheon",
"Michael J.",
""
],
[
"Hansen",
"Mads F.",
""
],
[
"Dolui",
"Kapildeb",
""
],
[
"Sinha",
"Shubham",
""
],
[
"Sahoo",
"Mihir R.",
""
],
[
"Pickard",
"Chris J.",
""
],
[
"Heil",
"Christoph",
""
],
[
"Pakhomova",
"Anna",
""
],
[
"Mezouar",
"Mohamed",
""
],
[
"Daisenberger",
"Dominik",
""
],
[
"Chariton",
"Stella",
""
],
[
"Prakapenka",
"Vitali",
""
],
[
"Julian",
"Matthew N.",
""
],
[
"Prasankumar",
"Rohit P.",
""
],
[
"Strobel",
"Timothy A.",
""
]
] |
2505.22547 | LiCheng Feng | T2K Collaboration | First measurement of neutron capture multiplicity in neutrino-oxygen neutral-current quasi-elastic-like interactions using an accelerator neutrino beam | 27 pages, 20 figures | null | null | null | hep-ex physics.ins-det | http://creativecommons.org/licenses/by/4.0/ | We report the first measurement of neutron capture multiplicity in neutrino-oxygen neutral-current quasi-elastic-like interactions at the gadolinium-loaded Super-Kamiokande detector using the T2K neutrino beam, which has a peak energy of about 0.6 GeV. A total of 30 neutral-current quasi-elastic-like event candidates were selected from T2K data corresponding to an exposure of $1.76\times10^{20}$ protons on target. The $\gamma$ ray signals resulting from neutron captures were identified using a neural network. The flux-averaged mean neutron capture multiplicity was measured to be $1.37\pm0.33\text{ (stat.)}$$(^{+0.17}_{-0.27})$, which is compatible within $2.3\,\sigma$ than predictions obtained using our nominal simulation. We discuss potential sources of systematic uncertainty in the prediction and demonstrate that a significant portion of this discrepancy arises from the modeling of hadron-nucleus interactions in the detector medium. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:32:06 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"T2K Collaboration",
"",
""
]
] |
2505.22548 | Changhao Song | Changhao Song, Yazhou Zhang and Peng Zhang | Emotion-o1: Adaptive Long Reasoning for Emotion Understanding in LLMs | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Emotion understanding includes basic tasks (e.g., sentiment/emotion classification) and advanced tasks (e.g., sarcasm/humor detection). Current methods rely on fixed-length CoT reasoning, failing to adapt to the varying complexity of emotions. We propose a task-adaptive reasoning framework that employs DeepSeek-R1 to generate variable-length reasoning chains for different emotion tasks. By combining fine-tuning with reinforcement learning, we design a composite reward function that balances four objectives: prediction accuracy, adaptive reasoning depth control, structural diversity in reasoning paths, and suppression of repetitive logic. This approach achieves dynamic context-sensitive inference while enabling LLMs to autonomously develop deep reasoning capabilities. Experimental results demonstrate consistent improvements in both Acc and F1 scores across four tasks: emotion, sentiment, humor, and sarcasm. Notably, peak enhancements reached 3.56% F1 (2.76% Acc) for basic tasks and 37.95% F1 (23.14% Acc) for advanced tasks. Our work bridges rigid CoT reasoning and emotional complexity through adaptive-depth analysis. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:32:16 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Song",
"Changhao",
""
],
[
"Zhang",
"Yazhou",
""
],
[
"Zhang",
"Peng",
""
]
] |
2505.22549 | Alex Iacob | Alex Iacob, Lorenzo Sani, Mher Safaryan, Paris Giampouras, Samuel Horv\'ath, Andrej Jovanovic, Meghdad Kurmanji, Preslav Aleksandrov, William F. Shen, Xinchi Qiu, Nicholas D. Lane | DES-LOC: Desynced Low Communication Adaptive Optimizers for Training Foundation Models | Keywords: Distributed Training, Foundation Models, Large Language Models, Optimizers, Communication Efficiency, Federated Learning, Distributed Systems, Optimization Theory, Scaling, Robustness. Preprint, under review at NeurIPS | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Scaling foundation model training with Distributed Data Parallel (DDP) methods is bandwidth-limited. Existing infrequent communication methods like Local SGD were designed to synchronize only model parameters and cannot be trivially applied to adaptive optimizers due to additional optimizer states. Current approaches extending Local SGD either lack convergence guarantees or require synchronizing all optimizer states, tripling communication costs. We propose Desynced Low Communication Adaptive Optimizers (DES-LOC), a family of optimizers assigning independent synchronization periods to parameters and momenta, enabling lower communication costs while preserving convergence. Through extensive experiments on language models of up to 1.7B, we show that DES-LOC can communicate 170x less than DDP and 2x less than the previous state-of-the-art Local ADAM. Furthermore, unlike previous heuristic approaches, DES-LOC is suited for practical training scenarios prone to system failures. DES-LOC offers a scalable, bandwidth-efficient, and fault-tolerant solution for foundation model training. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:32:33 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Iacob",
"Alex",
""
],
[
"Sani",
"Lorenzo",
""
],
[
"Safaryan",
"Mher",
""
],
[
"Giampouras",
"Paris",
""
],
[
"Horváth",
"Samuel",
""
],
[
"Jovanovic",
"Andrej",
""
],
[
"Kurmanji",
"Meghdad",
""
],
[
"Aleksandrov",
"Preslav",
""
],
[
"Shen",
"William F.",
""
],
[
"Qiu",
"Xinchi",
""
],
[
"Lane",
"Nicholas D.",
""
]
] |
2505.22550 | Atul Kumar | Rosario Uceda-Sosa, Nandana Mihindukulasooriya, Atul Kumar, Sahil Bansal, Seema Nagar | Domain specific ontologies from Linked Open Data (LOD) | null | null | null | null | cs.IR | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Logical and probabilistic reasoning tasks that require a deeper knowledge of semantics are increasingly relying on general purpose ontologies such as Wikidata and DBpedia. However, tasks such as entity disambiguation and linking may benefit from domain specific knowledge graphs, which make it more efficient to consume the knowledge and easier to extend with proprietary content. We discuss our experience bootstrapping one such ontology for IT with a domain-agnostic pipeline, and extending it using domain-specific glossaries. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:33:01 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Uceda-Sosa",
"Rosario",
""
],
[
"Mihindukulasooriya",
"Nandana",
""
],
[
"Kumar",
"Atul",
""
],
[
"Bansal",
"Sahil",
""
],
[
"Nagar",
"Seema",
""
]
] |
2505.22551 | Long Hui | Long Hui, Wai Lok Yeung | Deep Learning-Based BMD Estimation from Radiographs with Conformal Uncertainty Quantification | null | null | null | null | cs.CV stat.AP | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Limited DXA access hinders osteoporosis screening. This proof-of-concept study proposes using widely available knee X-rays for opportunistic Bone Mineral Density (BMD) estimation via deep learning, emphasizing robust uncertainty quantification essential for clinical use. An EfficientNet model was trained on the OAI dataset to predict BMD from bilateral knee radiographs. Two Test-Time Augmentation (TTA) methods were compared: traditional averaging and a multi-sample approach. Crucially, Split Conformal Prediction was implemented to provide statistically rigorous, patient-specific prediction intervals with guaranteed coverage. Results showed a Pearson correlation of 0.68 (traditional TTA). While traditional TTA yielded better point predictions, the multi-sample approach produced slightly tighter confidence intervals (90%, 95%, 99%) while maintaining coverage. The framework appropriately expressed higher uncertainty for challenging cases. Although anatomical mismatch between knee X-rays and standard DXA limits immediate clinical use, this method establishes a foundation for trustworthy AI-assisted BMD screening using routine radiographs, potentially improving early osteoporosis detection. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:33:49 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hui",
"Long",
""
],
[
"Yeung",
"Wai Lok",
""
]
] |
2505.22552 | Hoang Pham | Hoang Pham, Thanh-Do Nguyen and Khac-Hoai Nam Bui | ClaimPKG: Enhancing Claim Verification via Pseudo-Subgraph Generation with Lightweight Specialized LLM | Accepted by ACL 2025 findings | ACL 2025 | null | null | cs.CL cs.AI cs.DB | http://creativecommons.org/licenses/by/4.0/ | Integrating knowledge graphs (KGs) to enhance the reasoning capabilities of large language models (LLMs) is an emerging research challenge in claim verification. While KGs provide structured, semantically rich representations well-suited for reasoning, most existing verification methods rely on unstructured text corpora, limiting their ability to effectively leverage KGs. Additionally, despite possessing strong reasoning abilities, modern LLMs struggle with multi-step modular pipelines and reasoning over KGs without adaptation. To address these challenges, we propose ClaimPKG, an end-to-end framework that seamlessly integrates LLM reasoning with structured knowledge from KGs. Specifically, the main idea of ClaimPKG is to employ a lightweight, specialized LLM to represent the input claim as pseudo-subgraphs, guiding a dedicated subgraph retrieval module to identify relevant KG subgraphs. These retrieved subgraphs are then processed by a general-purpose LLM to produce the final verdict and justification. Extensive experiments on the FactKG dataset demonstrate that ClaimPKG achieves state-of-the-art performance, outperforming strong baselines in this research field by 9%-12% accuracy points across multiple categories. Furthermore, ClaimPKG exhibits zero-shot generalizability to unstructured datasets such as HoVer and FEVEROUS, effectively combining structured knowledge from KGs with LLM reasoning across various LLM backbones. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:34:14 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Pham",
"Hoang",
""
],
[
"Nguyen",
"Thanh-Do",
""
],
[
"Bui",
"Khac-Hoai Nam",
""
]
] |
2505.22553 | Oem Trivedi | Jan Nov\'ak and Oem Trivedi | The old problem of the cosmological constant solved? | 18 pages with no figures, this essay was submitted for the 2025 Gravity Research Foundation Essay Award and comments are welcome! | null | null | null | gr-qc astro-ph.CO hep-th | http://creativecommons.org/licenses/by/4.0/ | We formulate an approach to quantum gravity, called the ring paradigm. Gravity is mediated superluminally, and the graviton is described as a phonon on the grid of matter in the Universe. This theory has very interesting applications to cosmology and would ultimately solve the old problem of the cosmological constant. It further gives new impulses to the scalar field theories because the gravitational ring decays to some phantom field. As is obvious, we radically break the Lorentz invariance, which means that some generalization of the Haag-Lopuszanski-Sohnius theorem in quantum field theory is possible. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:34:48 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Novák",
"Jan",
""
],
[
"Trivedi",
"Oem",
""
]
] |
2505.22554 | Agnideep Aich | Agnideep Aich, Md Monzur Murshed, Amanda Mayeaux, Sameera Hewage | Can Copulas Be Used for Feature Selection? A Machine Learning Study on Diabetes Risk Prediction | Submitted | null | null | null | stat.ML cs.LG | http://creativecommons.org/licenses/by/4.0/ | Accurate diabetes risk prediction relies on identifying key features from complex health datasets, but conventional methods like mutual information (MI) filters and genetic algorithms (GAs) often overlook extreme dependencies critical for high-risk subpopulations. In this study we introduce a feature-selection framework using the upper-tail dependence coefficient ({\lambda}U) of the novel A2 copula, which quantifies how often extreme higher values of a predictor co-occur with diabetes diagnoses (target variable). Applied to the CDC Diabetes Health Indicators dataset (n=253,680), our method prioritizes five predictors (self-reported general health, high blood pressure, body mass index, mobility limitations, and high cholesterol levels) based on upper tail dependencies. These features match or outperform MI and GA selected subsets across four classifiers (Random Forest, XGBoost, Logistic Regression, Gradient Boosting), achieving accuracy up to 86.5% (XGBoost) and AUC up to 0.806 (Gradient Boosting), rivaling the full 21-feature model. Permutation importance confirms clinical relevance, with BMI and general health driving accuracy. To our knowledge, this is the first work to apply a copula's upper-tail dependence for supervised feature selection, bridging extreme-value theory and machine learning to deliver a practical toolkit for diabetes prevention. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:34:58 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Aich",
"Agnideep",
""
],
[
"Murshed",
"Md Monzur",
""
],
[
"Mayeaux",
"Amanda",
""
],
[
"Hewage",
"Sameera",
""
]
] |
2505.22555 | Yanyi Qu | Yanyi Qu, Haoyang Ma, and Wenhui Xiong | MultiFormer: A Multi-Person Pose Estimation System Based on CSI and Attention Mechanism | null | null | null | null | cs.CV eess.SP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Human pose estimation based on Channel State Information (CSI) has emerged as a promising approach for non-intrusive and precise human activity monitoring, yet faces challenges including accurate multi-person pose recognition and effective CSI feature learning. This paper presents MultiFormer, a wireless sensing system that accurately estimates human pose through CSI. The proposed system adopts a Transformer based time-frequency dual-token feature extractor with multi-head self-attention. This feature extractor is able to model inter-subcarrier correlations and temporal dependencies of the CSI. The extracted CSI features and the pose probability heatmaps are then fused by Multi-Stage Feature Fusion Network (MSFN) to enforce the anatomical constraints. Extensive experiments conducted on on the public MM-Fi dataset and our self-collected dataset show that the MultiFormer achieves higher accuracy over state-of-the-art approaches, especially for high-mobility keypoints (wrists, elbows) that are particularly difficult for previous methods to accurately estimate. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:36:02 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Qu",
"Yanyi",
""
],
[
"Ma",
"Haoyang",
""
],
[
"Xiong",
"Wenhui",
""
]
] |
2505.22556 | Anton Lukyanenko | Brandon G. Barreto-Rosa, Jean-Philippe Burelle, Anton Lukyanenko, Martha Richey | Products of exact dynamical systems and Lorentzian continued fractions | 15 pages | null | null | null | math.DS math.NT | http://creativecommons.org/licenses/by/4.0/ | We describe a new continued fraction system in Minkowski space $\mathbb R^{1,1}$, proving convergence, ergodicity with respect to an explicit invariant measure, and Lagrange's theorem. The proof of ergodicity leads us to the question of exactness for products of dynamical systems. Under technical assumptions, namely Renyi's condition, we show that products of exact dynamical systems are again exact, allowing us to study $\alpha$-type perturbations of the system. In addition, we describe new CF systems in $\mathbb R^{1,1}$ and $\mathbb R^{2,1}\cong \mathrm{Sym}_2(\mathbb R)$ that, based on experimental evidence, we conjecture to be convergent and ergodic with respect to a finite invariant measure. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:36:59 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Barreto-Rosa",
"Brandon G.",
""
],
[
"Burelle",
"Jean-Philippe",
""
],
[
"Lukyanenko",
"Anton",
""
],
[
"Richey",
"Martha",
""
]
] |
2505.22557 | Valery Alexeev | Valery Alexeev and Philip Engel | On lattice-polarized K3 surfaces | null | null | null | null | math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We correct the commonly used definitions of lattice-polarized and lattice-quasipolarized K3 surfaces. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:38:02 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Alexeev",
"Valery",
""
],
[
"Engel",
"Philip",
""
]
] |
2505.22558 | Anatoly Galikhanov | Anatoly Galikhanov | Homotopical Observables and the Langlands Program via $\infty$-Topoi | 23 pages, 1 figure, 5 tables | null | null | null | math.NT math.AG math.AT math.CT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a pro-\'etale geometric object $D_\infty$ arising naturally from the tower of Artin-Schreier extensions in characteristic 2, equipped with a canonical endofunctor $O$ whose fixed points correspond to automorphic representations of $\mathrm{GL}_2(\mathbb{A}_{\mathbb{F}_2})$. The main theorem establishes that invariant predicates on $D_\infty$ parametrize cuspidal automorphic representations, preserving $L$-functions. We provide complete proofs using $\infty$-categorical techniques, explicit computations for small cases, and establish connections to discrete conformal field theory. As applications, we resolve the Carlitz-Drinfeld uniformization conjecture for function fields and compute previously unknown motivic cohomology groups. Our approach differs fundamentally from coalgebraic models by working internally in topoi and connecting to arithmetic geometry. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:38:22 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Galikhanov",
"Anatoly",
""
]
] |
2505.22559 | Xiaoyang Zhu | Jeongheon Choe, Taketo Handa, Chun-Ying Huang, Andr\'e Koch Liston, Jordan Cox, Jonathan Stensberg, Yongseok Hong, Daniel G. Chica, Ding Xu, Fuyang Tay, Vinicius da Silveira Lanza Avelar, Eric A. Arsenault, James McIver, Dmitri N. Basov, Milan Delor, Xavier Roy, X.-Y. Zhu | Observation of Coherent Ferrons | 39 pages, 4 figures, 10 supporting figures | null | null | null | cond-mat.mtrl-sci physics.app-ph physics.optics | http://creativecommons.org/licenses/by/4.0/ | Excitation of ordered phases produces quasiparticles and collective modes, as exemplified by magnons that emerge from magnetic order, with applications in information transmission and quantum interconnects. Extending this paradigm to ferroelectric materials suggests the existence of ferrons, i.e. fundamental quanta of the collective excitation of ferroelectric order5 developed theoretically by Bauer and coworkers. While coherent magnons are observed in a broad range of experiments, coherent ferrons have eluded experimental detection. This discrepancy is particularly intriguing given that electric dipole interactions (FE) are inherently stronger than their magnetic counterparts. Here, we report the generation and transport of coherent ferrons in the van der Waals (vdW) ferroelectric material NbOI2. By launching collective oscillations of the ferroelectric dipoles using a short laser pulse, we identify coherent ferrons from intense and narrow-band terahertz (THz) emission and observe their propagations along the polar direction at extremely hypersonic velocities exceeding 10^5 m/s. The THz emission is a second-order nonlinear process that requires ferroelectric order, as is confirmed in the structurally related ferroelectric WO2Br2 and non-ferroelectric TaOBr2. The discovery of coherent ferrons paves the way for numerous applications, including narrow-band THz emission, ferronic information processing, and quantum interconnects. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:38:26 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Choe",
"Jeongheon",
""
],
[
"Handa",
"Taketo",
""
],
[
"Huang",
"Chun-Ying",
""
],
[
"Liston",
"André Koch",
""
],
[
"Cox",
"Jordan",
""
],
[
"Stensberg",
"Jonathan",
""
],
[
"Hong",
"Yongseok",
""
],
[
"Chica",
"Daniel G.",
""
],
[
"Xu",
"Ding",
""
],
[
"Tay",
"Fuyang",
""
],
[
"Avelar",
"Vinicius da Silveira Lanza",
""
],
[
"Arsenault",
"Eric A.",
""
],
[
"McIver",
"James",
""
],
[
"Basov",
"Dmitri N.",
""
],
[
"Delor",
"Milan",
""
],
[
"Roy",
"Xavier",
""
],
[
"Zhu",
"X. -Y.",
""
]
] |
2505.22560 | Artem Moskalev | Artem Moskalev, Mangal Prakash, Junjie Xu, Tianyu Cui, Rui Liao, Tommaso Mansi | Geometric Hyena Networks for Large-scale Equivariant Learning | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Processing global geometric context while preserving equivariance is crucial when modeling biological, chemical, and physical systems. Yet, this is challenging due to the computational demands of equivariance and global context at scale. Standard methods such as equivariant self-attention suffer from quadratic complexity, while local methods such as distance-based message passing sacrifice global information. Inspired by the recent success of state-space and long-convolutional models, we introduce Geometric Hyena, the first equivariant long-convolutional model for geometric systems. Geometric Hyena captures global geometric context at sub-quadratic complexity while maintaining equivariance to rotations and translations. Evaluated on all-atom property prediction of large RNA molecules and full protein molecular dynamics, Geometric Hyena outperforms existing equivariant models while requiring significantly less memory and compute that equivariant self-attention. Notably, our model processes the geometric context of 30k tokens 20x faster than the equivariant transformer and allows 72x longer context within the same budget. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:38:35 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Moskalev",
"Artem",
""
],
[
"Prakash",
"Mangal",
""
],
[
"Xu",
"Junjie",
""
],
[
"Cui",
"Tianyu",
""
],
[
"Liao",
"Rui",
""
],
[
"Mansi",
"Tommaso",
""
]
] |
2505.22561 | Jan Hubi\v{c}ka | Jan Hubi\v{c}ka, Mat\v{e}j Kone\v{c}n\'y, Stevo Todorcevic, Andy Zucker | On Big Ramsey degrees of universal $\omega$-edge-labeled hypergraphs | 6 pages, 1 figure, accepted to Eurocomb 2025 | null | null | null | math.CO cs.DM math.LO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that the big Ramsey degrees of every countable universal $u$-uniform $\omega$-edge-labeled hypergraph are infinite for every $u\geq 2$. Together with a recent result of Braunfeld, Chodounsk\'y, de Rancourt, Hubi\v{c}ka, Kawach, and Kone\v{c}n\'y this finishes full characterisation of unrestricted relational structures with finite big Ramsey degrees. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:38:40 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hubička",
"Jan",
""
],
[
"Konečný",
"Matěj",
""
],
[
"Todorcevic",
"Stevo",
""
],
[
"Zucker",
"Andy",
""
]
] |
2505.22563 | Yu Lei | Yu Lei, Xingyang Ge, Yi Zhang, Yiming Yang, Bolei Ma | Do Large Language Models Think Like the Brain? Sentence-Level Evidence from fMRI and Hierarchical Embeddings | null | null | null | null | cs.CL q-bio.NC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Understanding whether large language models (LLMs) and the human brain converge on similar computational principles remains a fundamental and important question in cognitive neuroscience and AI. Do the brain-like patterns observed in LLMs emerge simply from scaling, or do they reflect deeper alignment with the architecture of human language processing? This study focuses on the sentence-level neural mechanisms of language models, systematically investigating how hierarchical representations in LLMs align with the dynamic neural responses during human sentence comprehension. By comparing hierarchical embeddings from 14 publicly available LLMs with fMRI data collected from participants, who were exposed to a naturalistic narrative story, we constructed sentence-level neural prediction models to precisely identify the model layers most significantly correlated with brain region activations. Results show that improvements in model performance drive the evolution of representational architectures toward brain-like hierarchies, particularly achieving stronger functional and anatomical correspondence at higher semantic abstraction levels. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:40:06 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lei",
"Yu",
""
],
[
"Ge",
"Xingyang",
""
],
[
"Zhang",
"Yi",
""
],
[
"Yang",
"Yiming",
""
],
[
"Ma",
"Bolei",
""
]
] |
2505.22564 | Jaehyun Choi | Jaehyun Choi, Jiwan Hur, Gyojin Han, Jaemyung Yu, Junmo Kim | PRISM: Video Dataset Condensation with Progressive Refinement and Insertion for Sparse Motion | null | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Video dataset condensation has emerged as a critical technique for addressing the computational challenges associated with large-scale video data processing in deep learning applications. While significant progress has been made in image dataset condensation, the video domain presents unique challenges due to the complex interplay between spatial content and temporal dynamics. This paper introduces PRISM, Progressive Refinement and Insertion for Sparse Motion, for video dataset condensation, a novel approach that fundamentally reconsiders how video data should be condensed. Unlike the previous method that separates static content from dynamic motion, our method preserves the essential interdependence between these elements. Our approach progressively refines and inserts frames to fully accommodate the motion in an action while achieving better performance but less storage, considering the relation of gradients for each frame. Extensive experiments across standard video action recognition benchmarks demonstrate that PRISM outperforms existing disentangled approaches while maintaining compact representations suitable for resource-constrained environments. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:42:10 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Choi",
"Jaehyun",
""
],
[
"Hur",
"Jiwan",
""
],
[
"Han",
"Gyojin",
""
],
[
"Yu",
"Jaemyung",
""
],
[
"Kim",
"Junmo",
""
]
] |
2505.22565 | David Craven | David A. Craven | The Ingleton inequality holds for metacyclic groups and fails for supersoluble groups | 14 pages | null | null | null | math.GR cs.IT math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Ingleton inequality first appeared in matroid theory, where Ingleton proved in 1971 that every rank function coming from a representable matroid on four subsets satisfies a particular inequality. Because this inequality is not implied by submodularity, Shannon-type axioms alone, it and various analogues play a central role in separately linear and non-linear phenomena in a variety of areas of mathematics. The Ingleton inequality for finite groups concerns the various intersections of four subgroups. It holds for many quadruples of subgroups of finite groups, but not all, the smallest example being four subgroups of $S_5$, of order 120. Open questions are whether the Inlgeton inequality always holds for metacycle and nilpotent groups. (There is a proof in the literature due to Oggier and Stancu, but there is an already known issue with their proof, which we address in this article.)
In this paper we prove that the Ingleton inequality always holds for metacycle groups, but that it fails for supersoluble groups, a class of groups only a little larger than nilpotent groups. Although we do not resolve the nilpotent case here we do make some reductions, and also prove that there are no nilpotent violators of the Ingleton inequality of order less than 1024. We end with a list of Ingleton inequality violating groups of order at most 1023.
The article comes with a Magma package that allows reproduction of all results in the paper and for the reader to check the Ingleton inequality for any given finite group. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:42:48 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Craven",
"David A.",
""
]
] |
2505.22566 | Yifan Xie | Yifan Xie, Mingyang Li, Shoujie Li, Xingting Li, Guangyu Chen, Fei Ma, Fei Richard Yu, Wenbo Ding | Universal Visuo-Tactile Video Understanding for Embodied Interaction | 13 pages, 5 figures | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Tactile perception is essential for embodied agents to understand physical attributes of objects that cannot be determined through visual inspection alone. While existing approaches have made progress in visual and language modalities for physical understanding, they fail to effectively incorporate tactile information that provides crucial haptic feedback for real-world interaction. In this paper, we present VTV-LLM, the first multi-modal large language model for universal Visuo-Tactile Video (VTV) understanding that bridges the gap between tactile perception and natural language. To address the challenges of cross-sensor and cross-modal integration, we contribute VTV150K, a comprehensive dataset comprising 150,000 video frames from 100 diverse objects captured across three different tactile sensors (GelSight Mini, DIGIT, and Tac3D), annotated with four fundamental tactile attributes (hardness, protrusion, elasticity, and friction). We develop a novel three-stage training paradigm that includes VTV enhancement for robust visuo-tactile representation, VTV-text alignment for cross-modal correspondence, and text prompt finetuning for natural language generation. Our framework enables sophisticated tactile reasoning capabilities including feature assessment, comparative analysis, scenario-based decision making and so on. Experimental evaluations demonstrate that VTV-LLM achieves superior performance in tactile video understanding tasks, establishing a foundation for more intuitive human-machine interaction in tactile domains. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:43:01 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Xie",
"Yifan",
""
],
[
"Li",
"Mingyang",
""
],
[
"Li",
"Shoujie",
""
],
[
"Li",
"Xingting",
""
],
[
"Chen",
"Guangyu",
""
],
[
"Ma",
"Fei",
""
],
[
"Yu",
"Fei Richard",
""
],
[
"Ding",
"Wenbo",
""
]
] |
2505.22568 | Aravind Ramkumar Krishnan | Aravind R. Krishnan, Thomas Z. Li, Lucas W. Remedios, Michael E. Kim, Chenyu Gao, Gaurav Rudravaram, Elyssa M. McMaster, Adam M. Saunders, Shunxing Bao, Kaiwen Xu, Lianrui Zuo, Kim L. Sandler, Fabien Maldonado, Yuankai Huo, Bennett A. Landman | Multipath cycleGAN for harmonization of paired and unpaired low-dose lung computed tomography reconstruction kernels | null | null | null | null | eess.IV cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Reconstruction kernels in computed tomography (CT) affect spatial resolution and noise characteristics, introducing systematic variability in quantitative imaging measurements such as emphysema quantification. Choosing an appropriate kernel is therefore essential for consistent quantitative analysis. We propose a multipath cycleGAN model for CT kernel harmonization, trained on a mixture of paired and unpaired data from a low-dose lung cancer screening cohort. The model features domain-specific encoders and decoders with a shared latent space and uses discriminators tailored for each domain.We train the model on 42 kernel combinations using 100 scans each from seven representative kernels in the National Lung Screening Trial (NLST) dataset. To evaluate performance, 240 scans from each kernel are harmonized to a reference soft kernel, and emphysema is quantified before and after harmonization. A general linear model assesses the impact of age, sex, smoking status, and kernel on emphysema. We also evaluate harmonization from soft kernels to a reference hard kernel. To assess anatomical consistency, we compare segmentations of lung vessels, muscle, and subcutaneous adipose tissue generated by TotalSegmentator between harmonized and original images. Our model is benchmarked against traditional and switchable cycleGANs. For paired kernels, our approach reduces bias in emphysema scores, as seen in Bland-Altman plots (p<0.05). For unpaired kernels, harmonization eliminates confounding differences in emphysema (p>0.05). High Dice scores confirm preservation of muscle and fat anatomy, while lung vessel overlap remains reasonable. Overall, our shared latent space multipath cycleGAN enables robust harmonization across paired and unpaired CT kernels, improving emphysema quantification and preserving anatomical fidelity. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:44:42 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Krishnan",
"Aravind R.",
""
],
[
"Li",
"Thomas Z.",
""
],
[
"Remedios",
"Lucas W.",
""
],
[
"Kim",
"Michael E.",
""
],
[
"Gao",
"Chenyu",
""
],
[
"Rudravaram",
"Gaurav",
""
],
[
"McMaster",
"Elyssa M.",
""
],
[
"Saunders",
"Adam M.",
""
],
[
"Bao",
"Shunxing",
""
],
[
"Xu",
"Kaiwen",
""
],
[
"Zuo",
"Lianrui",
""
],
[
"Sandler",
"Kim L.",
""
],
[
"Maldonado",
"Fabien",
""
],
[
"Huo",
"Yuankai",
""
],
[
"Landman",
"Bennett A.",
""
]
] |
2505.22569 | Maksim Nakhodnov | Dmitrii Sorokin, Maksim Nakhodnov, Andrey Kuznetsov, Aibek Alanov | ImageReFL: Balancing Quality and Diversity in Human-Aligned Diffusion Models | The source code can be found at https://github.com/ControlGenAI/ImageReFL | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advances in diffusion models have led to impressive image generation capabilities, but aligning these models with human preferences remains challenging. Reward-based fine-tuning using models trained on human feedback improves alignment but often harms diversity, producing less varied outputs. In this work, we address this trade-off with two contributions. First, we introduce \textit{combined generation}, a novel sampling strategy that applies a reward-tuned diffusion model only in the later stages of the generation process, while preserving the base model for earlier steps. This approach mitigates early-stage overfitting and helps retain global structure and diversity. Second, we propose \textit{ImageReFL}, a fine-tuning method that improves image diversity with minimal loss in quality by training on real images and incorporating multiple regularizers, including diffusion and ReFL losses. Our approach outperforms conventional reward tuning methods on standard quality and diversity metrics. A user study further confirms that our method better balances human preference alignment and visual diversity. The source code can be found at https://github.com/ControlGenAI/ImageReFL . | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:45:07 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Sorokin",
"Dmitrii",
""
],
[
"Nakhodnov",
"Maksim",
""
],
[
"Kuznetsov",
"Andrey",
""
],
[
"Alanov",
"Aibek",
""
]
] |
2505.22570 | Maya Thompson | Iain Moffatt, Maya Thompson | Tensor product formulas for the Bollob\'as-Riordan and Krushkal polynomials | 27 pages, 11 figures | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | Brylawski's tensor product formula expresses the Tutte polynomial of the tensor product of two graphs in terms of Tutte polynomials arising from the tensor factors. Analogous tensor product formulas are known for the ribbon graph polynomial and transition polynomials of graphs embedded in surfaces, as well as for the Bollob\'as-Riordan polynomial in some special cases. We define the tensor product of graphs embedded in pseudo-surfaces and use this to generalize and unify all of the above results, providing Brylawski-style formulas for both the Bollob\'as-Riordan and Krushkal polynomials. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:45:20 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Moffatt",
"Iain",
""
],
[
"Thompson",
"Maya",
""
]
] |
2505.22572 | Waldemar Chang | Waldemar Chang, Alhassan Yasin | Fusion Steering: Prompt-Specific Activation Control | 14 pages, 4 figures, 2 tables | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | We present Fusion Steering, an activation steering methodology that improves factual accuracy in large language models (LLMs) for question-answering (QA) tasks. This approach introduces flexible steering configurations, including full-layer steering and segmented steering. Unlike traditional methods constrained to single-layer or fixed-layer operations, Fusion Steering employs dynamic injection of prompt-specific activation deltas across all transformer layers. These activation deltas are derived from reference completions that combine the ground-truth answer with a model-generated explanation to facilitate semantically enriched, example-specific steering. The injection weights are optimized per prompt using Optuna, targeting a joint objective that balances token overlap (factual alignment) and perplexity (fluency proxy). Evaluation employs a composite score integrating token overlap and LLM-graded quality, encompassing factual accuracy, coherence, and relevance. Empirical results on 260 SimpleQA prompts (selected from 500 where the baseline failed) showcase the efficacy of segmented steering. Using Gemma-2-2B-IT with 8-bit quantization, segmented steering achieves an accuracy of 25.4% (outputs scoring $\geq 0.6$), outperforming the baseline at 3.5% and full-layer steering at 16.2%. Under the stricter SimpleQA rubric, segmented steering boosts fully correct responses from 0.0% to 13.1%. These findings highlight the strengths of segmented, dynamic intervention strategies and the promise of per-prompt, full-network activation control. Fusion Steering is also amenable to sparse representations, such as Neuronpedia or sparse crosscoders, suggesting a promising direction for interpretable and scalable activation-level control in LLMs. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:46:55 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Chang",
"Waldemar",
""
],
[
"Yasin",
"Alhassan",
""
]
] |
2505.22573 | Guy Moss | Guy Moss, Leah Sophie Muhle, Reinhard Drews, Jakob H. Macke, Cornelius Schr\"oder | FNOPE: Simulation-based inference on function spaces with Fourier Neural Operators | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Simulation-based inference (SBI) is an established approach for performing Bayesian inference on scientific simulators. SBI so far works best on low-dimensional parametric models. However, it is difficult to infer function-valued parameters, which frequently occur in disciplines that model spatiotemporal processes such as the climate and earth sciences. Here, we introduce an approach for efficient posterior estimation, using a Fourier Neural Operator (FNO) architecture with a flow matching objective. We show that our approach, FNOPE, can perform inference of function-valued parameters at a fraction of the simulation budget of state of the art methods. In addition, FNOPE supports posterior evaluation at arbitrary discretizations of the domain, as well as simultaneous estimation of vector-valued parameters. We demonstrate the effectiveness of our approach on several benchmark tasks and a challenging spatial inference task from glaciology. FNOPE extends the applicability of SBI methods to new scientific domains by enabling the inference of function-valued parameters. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:46:56 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Moss",
"Guy",
""
],
[
"Muhle",
"Leah Sophie",
""
],
[
"Drews",
"Reinhard",
""
],
[
"Macke",
"Jakob H.",
""
],
[
"Schröder",
"Cornelius",
""
]
] |
2505.22574 | Yijie Zhu | Yijie Zhu, Evan Saraivanov, Joshua A. Kable, Artemis Sofia Giannakopoulou, Amritpal Nijjar, Vivian Miranda, Marco Bonici, Tim Eifler, and Elisabeth Krause | Attention-based Neural Network Emulators for Multi-Probe Data Vectors Part III: Modeling The Next Generation Surveys | 27 pages, 24 figures, 11 tables | null | null | null | astro-ph.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Machine learning can accelerate cosmological inferences that involve many sequential evaluations of computationally expensive data vectors. Previous works in this series have examined how machine learning architectures impact emulator accuracy and training time for optical shear and galaxy clustering 2-point function. In this final manuscript, we explore neural network performance when emulating Cosmic Microwave Background temperature and polarization power spectra. We maximize the volume of applicability in the parameter space of our emulators within the standard $\Lambda$-cold-dark-matter model while ensuring that errors are below cosmic variance. Relative to standard multi-layer perceptron architectures, we find the dot-product-attention mechanism reduces the number of outliers among testing cosmologies, defined as the fraction of testing points with $\Delta \chi^2 > 0.2$ relative to \textsc{CAMB} outputs, for a wide range of training set sizes. Such precision enables attention-based emulators to be directly applied to real data without requiring any additional correction via importance sampling. Combined with pre-processing techniques and optimized activation and loss functions, attention-based models can meet the precision criteria set by current and future CMB and lensing experiments. For each of Planck, Simons Observatory, CMB S4, and CMB HD, we find the fraction of outlier points to be less than $10\%$ with around $2\times10^5$ to $4\times10^5$ training data vectors. We further explore the applications of these methods to supernova distance, weak lensing, and galaxy clustering, as well as alternative architectures and pre-processing techniques. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:47:02 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhu",
"Yijie",
""
],
[
"Saraivanov",
"Evan",
""
],
[
"Kable",
"Joshua A.",
""
],
[
"Giannakopoulou",
"Artemis Sofia",
""
],
[
"Nijjar",
"Amritpal",
""
],
[
"Miranda",
"Vivian",
""
],
[
"Bonici",
"Marco",
""
],
[
"Eifler",
"Tim",
""
],
[
"Krause",
"Elisabeth",
""
]
] |
2505.22575 | Gerard McCaul | Gerard McCaul, Juan Sebastian Totero Gongora, Wendy Otieno, Sergey Savelev, Alexandre Zagoskin, and Alexander G. Balanov | Minimal Quantum Reservoirs with Hamiltonian Encoding | 13 pages, 7 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | We investigate a minimal architecture for quantum reservoir computing based on Hamiltonian encoding, in which input data is injected via modulation of system parameters rather than state preparation. This approach circumvents many of the experimental overheads typically associated with quantum machine learning, enabling computation without feedback, memory, or state tomography. We demonstrate that such a minimal quantum reservoir, despite lacking intrinsic memory, can perform nonlinear regression and prediction tasks when augmented with post-processing delay embeddings. Our results provide a conceptually and practically streamlined framework for quantum information processing, offering a clear baseline for future implementations on near-term quantum hardware. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:50:05 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"McCaul",
"Gerard",
""
],
[
"Gongora",
"Juan Sebastian Totero",
""
],
[
"Otieno",
"Wendy",
""
],
[
"Savelev",
"Sergey",
""
],
[
"Zagoskin",
"Alexandre",
""
],
[
"Balanov",
"Alexander G.",
""
]
] |
2505.22576 | Qing-He Ni | Qing-He Ni, Christian Hill, Sergei N. Yurchenko, Marco Pezzella, Alexander Fateev, Zhi Qin, Olivia Venot and Jonathan Tennyson | ExoPhoto: A Database of Temperature-Dependent Photodissociation Cross Sections | RAS Techniques and Instruments "in press" | null | null | null | astro-ph.EP astro-ph.IM physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | We present the ExoPhoto database (https://exomol.com/exophoto/), an extension of the ExoMol database, specifically developed to address the growing need for high-accuracy, temperature-dependent photodissociation cross section data towards short-UV wavelengths. ExoPhoto combines theoretical models from three major computational databases (ExoMol, UGAMOP and PhoMol) and experimental datasets from two experimental groups, providing extensive wavelength and temperature coverage. ExoPhoto currently includes photodissociation data for 20 molecules: AlH, HCl, HF, MgH, OH, NaO, MgO, O2, AlCl, AlF, CS, HeH+, CO, CO2, H2O, SO2, C2H2, C2H4, H2CO, and NH3, derived from theoretical models and supported by experimental data from 5 databases.
ExoPhoto also includes detailed data on branching ratios and quantum yields for selected datasets. The data structure of ExoPhoto follows the ExoMol framework, with a consistent naming convention and hierarchical JSON-based organization. Photodissociation cross sections are stored in a set of .photo files which provide data as a function of wavelength with one file for each target molecule temperature. Future developments aim to include more photodissociation cross section data and to provide data for molecules in non-local thermodynamic equilibrium (non-LTE). These will expand the utility of ExoPhoto for advanced astrophysical, planetary modeling and industrial applications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:51:17 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ni",
"Qing-He",
""
],
[
"Hill",
"Christian",
""
],
[
"Yurchenko",
"Sergei N.",
""
],
[
"Pezzella",
"Marco",
""
],
[
"Fateev",
"Alexander",
""
],
[
"Qin",
"Zhi",
""
],
[
"Venot",
"Olivia",
""
],
[
"Tennyson",
"Jonathan",
""
]
] |
2505.22577 | Elahe Khalili Samani | Elahe Khalili Samani, Marco Radeschi | Positive curvature and rational ellipticity in cohomogeneity three | null | null | null | null | math.DG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We prove that a closed, simply connected, positively curved, cohomogeneity-three manifold whose quotient space has no boundary is rationally elliptic, thus providing a generalization of similar results regarding rational ellipticity of homogeneous, cohomogeneity-one, and almost non-negatively curved cohomogeneity-two manifolds. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:51:35 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Samani",
"Elahe Khalili",
""
],
[
"Radeschi",
"Marco",
""
]
] |
2505.22578 | Etienne Boursier | Etienne Boursier, Matthew Bowditch, Matthias Englert, Ranko Lazic | Benignity of loss landscape with weight decay requires both large overparametrization and initialization | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The optimization of neural networks under weight decay remains poorly understood from a theoretical standpoint. While weight decay is standard practice in modern training procedures, most theoretical analyses focus on unregularized settings. In this work, we investigate the loss landscape of the $\ell_2$-regularized training loss for two-layer ReLU networks. We show that the landscape becomes benign -- i.e., free of spurious local minima -- under large overparametrization, specifically when the network width $m$ satisfies $m \gtrsim \min(n^d, 2^n)$, where $n$ is the number of data points and $d$ the input dimension. More precisely in this regime, almost all constant activation regions contain a global minimum and no spurious local minima. We further show that this level of overparametrization is not only sufficient but also necessary via the example of orthogonal data. Finally, we demonstrate that such loss landscape results primarily hold relevance in the large initialization regime. In contrast, for small initializations -- corresponding to the feature learning regime -- optimization can still converge to spurious local minima, despite the global benignity of the landscape. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:53:48 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Boursier",
"Etienne",
""
],
[
"Bowditch",
"Matthew",
""
],
[
"Englert",
"Matthias",
""
],
[
"Lazic",
"Ranko",
""
]
] |
2505.22579 | Reid Ferguson | Reid Ferguson, Olaf Hartwig, and Guido Mueller | Applying TDI to Hardware-Simulated Data | Contribution to the 2025 Gravitation session of the 59th Recontres de Moriond | null | null | null | gr-qc astro-ph.IM physics.ins-det | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Many years of development have gone into producing instruments that meet the required noise performance of the LISA interferometric detection system. Concurrently, software simulations have been used to extensively develop the data analysis libraries to be used in the LISA pipeline, not least among which are the Time Delay Interferometry (TDI) algorithms. To bridge the gap between these two, we are developing a hardware-in-the-loop testbed to apply realistic, time-varying delays to signals traveling between phasemeters. We have shown that the testbed adds a sufficiently low amount of noise across the entire relevant LISA spectrum. We have also injected a realistic gravitational wave signal, generated via the LISA Data Challenge codebase, and successfully extracted it using TDI to remove the obscuring frequency noise of the carrier signal. Future efforts will expand the testbed to create a representative simulation of the entire LISA constellation, with an eye towards its use as a tool to aid in the development of the LISA data analysis pipelines. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:54:25 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ferguson",
"Reid",
""
],
[
"Hartwig",
"Olaf",
""
],
[
"Mueller",
"Guido",
""
]
] |
2505.22580 | Louis Shuo Wang | Louis Shuo Wang, Jiguang Yu, Zonghao Liu | A hybrid PDE-ABM model for angiogenesis and tumour microenvironment with application to resistance in cancer treatment | null | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The main obstacle to effective cancer treatment is the development of drug resistance, which can be divided into two categories: spontaneous and acquired drug resistance. Non-small cell lung cancer (NSCLC) is the main cause of cancer-related deaths worldwide. A subset of lung cancer, adenocarcinomas, is characterised by mutations in the epidermal growth factor receptor (EGFR) gene. Treatment of EGFR-mutated lung adenocarcinomas has become less effective over time due to drug resistance development, which is associated with a second mutation in the EGFR gene. An important factor in the development of cancer is angiogenesis, which is the formation of blood vessels from the existing vasculature. These newly formed blood vessels provide oxygen and nutrients to tumour cells to maintain tumour growth and proliferation. We applied a hybrid discrete-continuous (HDC) model to capture the dynamic vasculature in the tumour microenvironment (TME). In the case of pre-existing resistance, the formation of angiogenic networks creates a microenvironment that supports tumour survival and enhances drug resistance. In the case of spontaneous mutation-induced resistance, earlier and more frequent mutations confer a greater survival advantage to the tumour population. There is also a mutually reinforcing relationship between a high proliferation rate and high resistance characteristics. These findings explain two conflicting experimental results about the second mutation in NSCLC. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:54:29 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wang",
"Louis Shuo",
""
],
[
"Yu",
"Jiguang",
""
],
[
"Liu",
"Zonghao",
""
]
] |
2505.22581 | Kartik Kuckreja | Kartik Kuckreja, Parul Gupta, Injy Hamed, Thamar Solorio, Muhammad Haris Khan, Abhinav Dhall | Tell me Habibi, is it Real or Fake? | 9 pages, 2 figures, 12 tables | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Deepfake generation methods are evolving fast, making fake media harder to detect and raising serious societal concerns. Most deepfake detection and dataset creation research focuses on monolingual content, often overlooking the challenges of multilingual and code-switched speech, where multiple languages are mixed within the same discourse. Code-switching, especially between Arabic and English, is common in the Arab world and is widely used in digital communication. This linguistic mixing poses extra challenges for deepfake detection, as it can confuse models trained mostly on monolingual data. To address this, we introduce \textbf{ArEnAV}, the first large-scale Arabic-English audio-visual deepfake dataset featuring intra-utterance code-switching, dialectal variation, and monolingual Arabic content. It \textbf{contains 387k videos and over 765 hours of real and fake videos}. Our dataset is generated using a novel pipeline integrating four Text-To-Speech and two lip-sync models, enabling comprehensive analysis of multilingual multimodal deepfake detection. We benchmark our dataset against existing monolingual and multilingual datasets, state-of-the-art deepfake detection models, and a human evaluation, highlighting its potential to advance deepfake research. The dataset can be accessed \href{https://huggingface.co/datasets/kartik060702/ArEnAV-Full}{here}. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:54:36 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Kuckreja",
"Kartik",
""
],
[
"Gupta",
"Parul",
""
],
[
"Hamed",
"Injy",
""
],
[
"Solorio",
"Thamar",
""
],
[
"Khan",
"Muhammad Haris",
""
],
[
"Dhall",
"Abhinav",
""
]
] |
2505.22582 | Xue Zhang | Xue Zhang, Yunlong Liang, Fandong Meng, Songming Zhang, Yufeng Chen, Jinan Xu, Jie Zhou | Less, but Better: Efficient Multilingual Expansion for LLMs via Layer-wise Mixture-of-Experts | ACL 2025 (Main), 16 pages, 5 figures, 11 tables | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Continually expanding new languages for existing large language models (LLMs) is a promising yet challenging approach to building powerful multilingual LLMs. The biggest challenge is to make the model continuously learn new languages while preserving the proficient ability of old languages. To achieve this, recent work utilizes the Mixture-of-Experts (MoE) architecture to expand new languages by adding new experts and avoid catastrophic forgetting of old languages by routing corresponding tokens to the original model backbone (old experts). Although intuitive, this kind of method is parameter-costly when expanding new languages and still inevitably impacts the performance of old languages. To address these limitations, we analyze the language characteristics of different layers in LLMs and propose a layer-wise expert allocation algorithm (LayerMoE) to determine the appropriate number of new experts for each layer. Specifically, we find different layers in LLMs exhibit different representation similarities between languages and then utilize the similarity as the indicator to allocate experts for each layer, i.e., the higher similarity, the fewer experts. Additionally, to further mitigate the forgetting of old languages, we add a classifier in front of the router network on the layers with higher similarity to guide the routing of old language tokens. Experimental results show that our method outperforms the previous state-of-the-art baseline with 60% fewer experts in the single-expansion setting and with 33.3% fewer experts in the lifelong-expansion setting, demonstrating the effectiveness of our method. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:54:53 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhang",
"Xue",
""
],
[
"Liang",
"Yunlong",
""
],
[
"Meng",
"Fandong",
""
],
[
"Zhang",
"Songming",
""
],
[
"Chen",
"Yufeng",
""
],
[
"Xu",
"Jinan",
""
],
[
"Zhou",
"Jie",
""
]
] |
2505.22583 | Tobias Lindenbauer | Tobias Lindenbauer, Egor Bogomolov, Yaroslav Zharov | GitGoodBench: A Novel Benchmark For Evaluating Agentic Performance On Git | Short Paper, 5 pages | null | null | null | cs.SE cs.AI | http://creativecommons.org/licenses/by-sa/4.0/ | Benchmarks for Software Engineering (SE) AI agents, most notably SWE-bench, have catalyzed progress in programming capabilities of AI agents. However, they overlook critical developer workflows such as Version Control System (VCS) operations. To address this issue, we present GitGoodBench, a novel benchmark for evaluating AI agent performance on VCS tasks. GitGoodBench covers three core Git scenarios extracted from permissive open-source Python, Java, and Kotlin repositories. Our benchmark provides three datasets: a comprehensive evaluation suite (900 samples), a rapid prototyping version (120 samples), and a training corpus (17,469 samples). We establish baseline performance on the prototyping version of our benchmark using GPT-4o equipped with custom tools, achieving a 21.11% solve rate overall. We expect GitGoodBench to serve as a crucial stepping stone toward truly comprehensive SE agents that go beyond mere programming. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:56:11 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lindenbauer",
"Tobias",
""
],
[
"Bogomolov",
"Egor",
""
],
[
"Zharov",
"Yaroslav",
""
]
] |
2505.22584 | Navve Wasserman | Navve Wasserman, Oliver Heinimann, Yuval Golbari, Tal Zimbalist, Eli Schwartz and Michal Irani | DocReRank: Single-Page Hard Negative Query Generation for Training Multi-Modal RAG Rerankers | null | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Rerankers play a critical role in multimodal Retrieval-Augmented Generation (RAG) by refining ranking of an initial set of retrieved documents. Rerankers are typically trained using hard negative mining, whose goal is to select pages for each query which rank high, but are actually irrelevant. However, this selection process is typically passive and restricted to what the retriever can find in the available corpus, leading to several inherent limitations. These include: limited diversity, negative examples which are often not hard enough, low controllability, and frequent false negatives which harm training. Our paper proposes an alternative approach: Single-Page Hard Negative Query Generation, which goes the other way around. Instead of retrieving negative pages per query, we generate hard negative queries per page. Using an automated LLM-VLM pipeline, and given a page and its positive query, we create hard negatives by rephrasing the query to be as similar as possible in form and context, yet not answerable from the page. This paradigm enables fine-grained control over the generated queries, resulting in diverse, hard, and targeted negatives. It also supports efficient false negative verification. Our experiments show that rerankers trained with data generated using our approach outperform existing models and significantly improve retrieval performance. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:56:41 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wasserman",
"Navve",
""
],
[
"Heinimann",
"Oliver",
""
],
[
"Golbari",
"Yuval",
""
],
[
"Zimbalist",
"Tal",
""
],
[
"Schwartz",
"Eli",
""
],
[
"Irani",
"Michal",
""
]
] |
2505.22585 | Jos\'e Nicol\'as Pi\~na Le\'on | N. Pi\~na-Le\'on, V. Manti\v{c}, S. Jim\'enez-Alfaro | A recursive method for computing singular solutions in corners with homogeneous Dirichlet-Robin boundary condition with power-law coefficient variation | null | null | null | null | math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study introduces a recursive method for computing asymptotic solutions of the Laplace equation in corner domains with the homogeneous Dirichlet boundary condition on one side and the Robin boundary condition with a power-law coefficient variation with exponent $\alpha\in \mathbb{R}$ on the other side (D-R corner problem). An asymptotic solution of this D-R corner problem is given as the sum of a main term, the solution of either a homogeneous Dirichlet-Neumann (D-N) or Dirichlet-Dirichlet (D-D) corner problem, and a finite or infinite series of the associated higher-order shadow terms by using harmonic basis functions with power-logarithmic terms. To determine this series of shadow terms, it is shown that the recursive procedures based on recursive non-homogeneous D-N or D-D corner problems are always convergent for $\alpha > -1$ or $\alpha < -1$, respectively. For the critical case $\alpha=-1$, the closed form expression of the asymptotic solution is given. Asymptotic solutions for several relevant D-R corner problems are derived and analysed. Two of these examples are applied to the problem of bridged cracks in antiplane Mode III in linear elastic fracture mechanics. The results presented can be applied to many other physical and engineering applications, such as heat transfer with the thermal resistance condition, acoustics and electrostatics with the impedance condition, and elasticity and structural analysis with the Winkler spring boundary condition. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:58:19 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Piña-León",
"N.",
""
],
[
"Mantič",
"V.",
""
],
[
"Jiménez-Alfaro",
"S.",
""
]
] |
2505.22586 | Yoav Gur-Arieh | Yoav Gur-Arieh, Clara Suslik, Yihuai Hong, Fazl Barez, Mor Geva | Precise In-Parameter Concept Erasure in Large Language Models | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Large language models (LLMs) often acquire knowledge during pretraining that is undesirable in downstream deployments, e.g., sensitive information or copyrighted content. Existing approaches for removing such knowledge rely on fine-tuning, training low-rank adapters or fact-level editing, but these are either too coarse, too shallow, or ineffective. In this work, we propose PISCES (Precise In-parameter Suppression for Concept EraSure), a novel framework for precisely erasing entire concepts from model parameters by directly editing directions that encode them in parameter space. PISCES uses a disentangler model to decompose MLP vectors into interpretable features, identifies those associated with a target concept using automated interpretability techniques, and removes them from model parameters. Experiments on Gemma 2 and Llama 3.1 over various concepts show that PISCES achieves modest gains in efficacy over leading erasure methods, reducing accuracy on the target concept to as low as 7.7%, while dramatically improving erasure specificity (by up to 31%) and robustness (by up to 38%). Overall, these results demonstrate that feature-based in-parameter editing enables a more precise and reliable approach for removing conceptual knowledge in language models. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:58:23 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Gur-Arieh",
"Yoav",
""
],
[
"Suslik",
"Clara",
""
],
[
"Hong",
"Yihuai",
""
],
[
"Barez",
"Fazl",
""
],
[
"Geva",
"Mor",
""
]
] |
2505.22587 | Zhenhua Lin | Bill Z. Lin, Simon Godsill | Bayesian Non-Parametric Inference for L\'evy Measures in State-Space Models | null | null | null | null | stat.ME | http://creativecommons.org/licenses/by/4.0/ | L\'evy processes, known for their ability to model complex dynamics with skewness, heavy tails and discontinuities, play a critical role in stochastic modeling across various domains. However, inference for most L\'evy processes, whether in parametric or non-parametric settings, remains a significant challenge. In this work, we present a novel Bayesian non-parametric inference framework for inferring the L\'evy measures of subordinators and normal variance-mean (NVM) processes within a linear L\'evy state space model, a setup that significantly extends existing methodologies. We employ the Dirichlet process which further results in a Student-t mixture representation to enable inference for the L\'evy measures. An efficient augmented Markov Chain Monte Carlo algorithm is developed for this problem that ensures both accuracy and computational feasibility. The effectiveness of the method is demonstrated on synthetic and tick-level (high-frequency) financial datasets, and we show the practical utility of the inference results using the forecasting performance. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:59:09 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lin",
"Bill Z.",
""
],
[
"Godsill",
"Simon",
""
]
] |
2505.22588 | Brian Hopkins | Brian Hopkins and James A. Sellers | Overpartitions and Kaur, Rana, and Eyyunni's mex sequences | 8 pages, 6 tables | null | null | null | math.CO math.NT | http://creativecommons.org/licenses/by/4.0/ | Kaur, Rana, and Eyyunni recently defined the mex sequence of a partition and established, by analytic methods, connections to two disparate types of partition-related objects. We make a bijection between partitions with certain mex sequences and a uniform family of overpartitions which allows us to provide combinatorial proofs of their results, as they requested. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:59:17 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hopkins",
"Brian",
""
],
[
"Sellers",
"James A.",
""
]
] |
2505.22589 | Alexey Litvinov | Alexey Bychkov and Alexey Litvinov | On dual regime in Yang-Baxter deformed $\mathrm{O}(2N)$ sigma models | null | null | null | null | hep-th math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | In this paper, we explore a new class of integrable sigma models, which we refer to as the "dual regime" of Yang-Baxter (YB) deformed $\mathrm{O}(2N)$ sigma models. This dual regime manifests itself in the conformal perturbation approach. Namely, it is well known that conventional YB-deformed $\mathrm{O}(N)$ sigma models are described in the UV by a collection of free bosonic fields perturbed by some relevant operators. The holomorphic parts of these operators play the role of screening operators which define certain integrable systems in the free theory. All of these integrable systems depend on a continuous parameter $b$, which parametrizes the central charge, and are known to possess the duality under $b^2\longleftrightarrow -1-b^2$. Although $\mathrm{O}(2N+1)$ integrable systems are self-dual, $\mathrm{O}(2N)$ systems are not. In particular, the $\mathrm{O}(2N)$ integrable systems provide new perturbations of the sigma model type. We identify the corresponding one-loop metric and $B-$field and show that they solve the generalized Ricci flow equation. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:59:24 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bychkov",
"Alexey",
""
],
[
"Litvinov",
"Alexey",
""
]
] |
2505.22590 | Paulo Luz | Paulo Luz and Filipe C. Mena | Local well-posedness of the initial value problem in Einstein-Cartan theory | 29 pages | null | null | null | gr-qc | http://creativecommons.org/licenses/by/4.0/ | We study the initial value problem in Einstein-Cartan theory which includes torsion and, therefore, a non-symmetric connection on the spacetime manifold. Generalizing the path of a classical theorem by Choquet-Bruhat and York for the Einstein equations, we use a $n+1$ splitting of the manifold and compute the evolution and constraint equations for the Einstein-Cartan system. In the process, we derive the Gauss-Codazzi-Ricci equations including torsion. We prove that the constraint equations are preserved during evolution. Imposing a generalized harmonic gauge, it is shown that the evolution equations can be cast as a quasilinear system in a Cauchy regular form with a characteristic determinant having a non-trivial multiplicity of characteristics. Using the Leray-Ohya theory for non-strictly hyperbolic systems we then establish the local geometric well-posedness of the Cauchy problem, for sufficiently regular initial data. For vanishing torsion we recover the classical results for the Einstein equations. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:00:04 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Luz",
"Paulo",
""
],
[
"Mena",
"Filipe C.",
""
]
] |
2505.22591 | Erxin Yu | Erxin Yu, Jing Li, Ming Liao, Qi Zhu, Boyang Xue, Minghui Xu, Baojun Wang, Lanqing Hong, Fei Mi, Lifeng Shang | Self-Error-Instruct: Generalizing from Errors for LLMs Mathematical Reasoning | 16 pages, 9 figures | null | null | null | cs.CL cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Although large language models demonstrate strong performance across various domains, they still struggle with numerous bad cases in mathematical reasoning. Previous approaches to learning from errors synthesize training data by solely extrapolating from isolated bad cases, thereby failing to generalize the extensive patterns inherent within these cases. This paper presents Self-Error-Instruct (SEI), a framework that addresses these model weaknesses and synthesizes more generalized targeted training data. Specifically, we explore a target model on two mathematical datasets, GSM8K and MATH, to pinpoint bad cases. Then, we generate error keyphrases for these cases based on the instructor model's (GPT-4o) analysis and identify error types by clustering these keyphrases. Next, we sample a few bad cases during each generation for each identified error type and input them into the instructor model, which synthesizes additional training data using a self-instruct approach. This new data is refined through a one-shot learning process to ensure that only the most effective examples are kept. Finally, we use these curated data to fine-tune the target model, iteratively repeating the process to enhance performance. We apply our framework to various models and observe improvements in their reasoning abilities across both in-domain and out-of-domain mathematics datasets. These results demonstrate the effectiveness of self-error instruction in improving LLMs' mathematical reasoning through error generalization. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:02:47 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Yu",
"Erxin",
""
],
[
"Li",
"Jing",
""
],
[
"Liao",
"Ming",
""
],
[
"Zhu",
"Qi",
""
],
[
"Xue",
"Boyang",
""
],
[
"Xu",
"Minghui",
""
],
[
"Wang",
"Baojun",
""
],
[
"Hong",
"Lanqing",
""
],
[
"Mi",
"Fei",
""
],
[
"Shang",
"Lifeng",
""
]
] |
2505.22592 | Yiheng Li | Yiheng Li, Francisco Carrillo-Perez, Mohammed Alawad, Olivier Gevaert | Comparative Analysis of Machine Learning Models for Lung Cancer Mutation Detection and Staging Using 3D CT Scans | null | null | null | null | eess.IV cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Lung cancer is the leading cause of cancer mortality worldwide, and non-invasive methods for detecting key mutations and staging are essential for improving patient outcomes. Here, we compare the performance of two machine learning models - FMCIB+XGBoost, a supervised model with domain-specific pretraining, and Dinov2+ABMIL, a self-supervised model with attention-based multiple-instance learning - on 3D lung nodule data from the Stanford Radiogenomics and Lung-CT-PT-Dx cohorts. In the task of KRAS and EGFR mutation detection, FMCIB+XGBoost consistently outperformed Dinov2+ABMIL, achieving accuracies of 0.846 and 0.883 for KRAS and EGFR mutations, respectively. In cancer staging, Dinov2+ABMIL demonstrated competitive generalization, achieving an accuracy of 0.797 for T-stage prediction in the Lung-CT-PT-Dx cohort, suggesting SSL's adaptability across diverse datasets. Our results emphasize the clinical utility of supervised models in mutation detection and highlight the potential of SSL to improve staging generalization, while identifying areas for enhancement in mutation sensitivity. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:04:35 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Li",
"Yiheng",
""
],
[
"Carrillo-Perez",
"Francisco",
""
],
[
"Alawad",
"Mohammed",
""
],
[
"Gevaert",
"Olivier",
""
]
] |
2505.22593 | Salah Elgendi | Nabil L. Youssef, Ebtsam H. Taha, A. A. Kotb and S. G. Elgendi | Special anisotropic conformal changes of conic pseudo-Finsler surfaces | 21 pages | null | null | null | math.DG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study presents many special anisotropic conformal changes of a conic pseudo-Finsler surface $(M,F)$, such as $C$-anisotropic and horizontal $C$-anisotropic conformal transformations, which reduce to $C$-conformal when the conformal factor is solely position-dependent. Furthermore, we present vertical $C$-anisotropic conformal changes and demonstrate that they are characterized by the property of $(M,F)$ being Riemannian. Additionally, we examine the anisotropic conformal transformation that fulfils the $\phi T$-condition, the horizontal $\phi T$-condition, and the vertical $\phi T$-condition. The first two conditions reduce to the $\boldsymbol{\sigma} T$-condition when the conformal factor relies solely on a positional variable. We demonstrate that, under the vertical $\phi T$-condition change, every Landsberg surface is Berwaldian. Thus, the vertical $\phi T$-condition is equivalent to the $T$-condition. Furthermore, we examine the scenario when the anisotropic conformal factor becomes the main scalar of the non-Riemannian surface $(M,F)$. We present an example of a Finslerian Schwarzschild-de Sitter solution having Finslerian spherical symmetry and apply our results to it. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:04:48 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Youssef",
"Nabil L.",
""
],
[
"Taha",
"Ebtsam H.",
""
],
[
"Kotb",
"A. A.",
""
],
[
"Elgendi",
"S. G.",
""
]
] |
2505.22594 | Yanke Song | Longlin Wang, Yanke Song, Kuanhao Jiang, Pragya Sur | GLAMP: An Approximate Message Passing Framework for Transfer Learning with Applications to Lasso-based Estimators | 104 pages, 3 figures | null | null | null | math.ST stat.ML stat.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Approximate Message Passing (AMP) algorithms enable precise characterization of certain classes of random objects in the high-dimensional limit, and have found widespread applications in fields such as statistics, deep learning, genetics, and communications. However, existing AMP frameworks cannot simultaneously handle matrix-valued iterates and non-separable denoising functions. This limitation prevents them from precisely characterizing estimators that draw information from multiple data sources with distribution shifts. In this work, we introduce Generalized Long Approximate Message Passing (GLAMP), a novel extension of AMP that addresses this limitation. We rigorously prove state evolution for GLAMP. GLAMP significantly broadens the scope of AMP, enabling the analysis of transfer learning estimators that were previously out of reach. We demonstrate the utility of GLAMP by precisely characterizing the risk of three Lasso-based transfer learning estimators: the Stacked Lasso, the Model Averaging Estimator, and the Second Step Estimator. We also demonstrate the remarkable finite sample accuracy of our theory via extensive simulations. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:05:09 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wang",
"Longlin",
""
],
[
"Song",
"Yanke",
""
],
[
"Jiang",
"Kuanhao",
""
],
[
"Sur",
"Pragya",
""
]
] |
2505.22595 | Andrei Belitsky | A.V. Belitsky, V.A. Smirnov | Off-shell form factor: factorization is violated | 31 pages, 3 figures | null | null | null | hep-th hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the Sudakov form factor on the Coulomb branch of N=4 sYM, which endows only external states with masses, and implies that the former is off-shell in the traditional sense. Our consideration is performed at three-loop order in the near mass-shell limit. We use a combination of tools to perform required calculations centered around the Method of Regions as the main go-to formalism for the asymptotic expansion of emerging parametric Feynman integrals. Explicit separation of quantum loops in terms of hard, collinear, and ultrasoft modes allows us to explore the factorization properties of this infrared-sensitive quantity. While the hard region is cleanly separated from the rest, the ultrasoft-collinear modes remain intertwined. We exhibit effects of factorization violation explicitly in the momentum space making use of the infrared power counting. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:07:01 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Belitsky",
"A. V.",
""
],
[
"Smirnov",
"V. A.",
""
]
] |
2505.22596 | Jiaqi Huang | Jiaqi Huang, Zunnan Xu, Jun Zhou, Ting Liu, Yicheng Xiao, Mingwen Ou, Bowen Ji, Xiu Li, Kehong Yuan | SAM-R1: Leveraging SAM for Reward Feedback in Multimodal Segmentation via Reinforcement Learning | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Leveraging multimodal large models for image segmentation has become a prominent research direction. However, existing approaches typically rely heavily on manually annotated datasets that include explicit reasoning processes, which are costly and time-consuming to produce. Recent advances suggest that reinforcement learning (RL) can endow large models with reasoning capabilities without requiring such reasoning-annotated data. In this paper, we propose SAM-R1, a novel framework that enables multimodal large models to perform fine-grained reasoning in image understanding tasks. Our approach is the first to incorporate fine-grained segmentation settings during the training of multimodal reasoning models. By integrating task-specific, fine-grained rewards with a tailored optimization objective, we further enhance the model's reasoning and segmentation alignment. We also leverage the Segment Anything Model (SAM) as a strong and flexible reward provider to guide the learning process. With only 3k training samples, SAM-R1 achieves strong performance across multiple benchmarks, demonstrating the effectiveness of reinforcement learning in equipping multimodal models with segmentation-oriented reasoning capabilities. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:08:28 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Huang",
"Jiaqi",
""
],
[
"Xu",
"Zunnan",
""
],
[
"Zhou",
"Jun",
""
],
[
"Liu",
"Ting",
""
],
[
"Xiao",
"Yicheng",
""
],
[
"Ou",
"Mingwen",
""
],
[
"Ji",
"Bowen",
""
],
[
"Li",
"Xiu",
""
],
[
"Yuan",
"Kehong",
""
]
] |
2505.22597 | Ngoc La | Ngoc La, Ruaridh Mon-Williams, and Julie A. Shah | HDDLGym: A Tool for Studying Multi-Agent Hierarchical Problems Defined in HDDL with OpenAI Gym | Accepted to Proceedings of ICAPS 2025 | null | null | null | cs.AI cs.LG cs.MA | http://creativecommons.org/licenses/by/4.0/ | In recent years, reinforcement learning (RL) methods have been widely tested using tools like OpenAI Gym, though many tasks in these environments could also benefit from hierarchical planning. However, there is a lack of a tool that enables seamless integration of hierarchical planning with RL. Hierarchical Domain Definition Language (HDDL), used in classical planning, introduces a structured approach well-suited for model-based RL to address this gap. To bridge this integration, we introduce HDDLGym, a Python-based tool that automatically generates OpenAI Gym environments from HDDL domains and problems. HDDLGym serves as a link between RL and hierarchical planning, supporting multi-agent scenarios and enabling collaborative planning among agents. This paper provides an overview of HDDLGym's design and implementation, highlighting the challenges and design choices involved in integrating HDDL with the Gym interface, and applying RL policies to support hierarchical planning. We also provide detailed instructions and demonstrations for using the HDDLGym framework, including how to work with existing HDDL domains and problems from International Planning Competitions, exemplified by the Transport domain. Additionally, we offer guidance on creating new HDDL domains for multi-agent scenarios and demonstrate the practical use of HDDLGym in the Overcooked domain. By leveraging the advantages of HDDL and Gym, HDDLGym aims to be a valuable tool for studying RL in hierarchical planning, particularly in multi-agent contexts. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:10:43 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"La",
"Ngoc",
""
],
[
"Mon-Williams",
"Ruaridh",
""
],
[
"Shah",
"Julie A.",
""
]
] |
2505.22599 | Hossein Rastgoftar | Jack T. Hughes, Mohammad Ghufran, and Hossein Rastgoftar | VR-Based Control of Multi-Copter Operation | null | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We aim to use virtual reality (VR) to improve the spatial awareness of pilots by real-time scanning of the environment around the drone using onboard sensors, live streaming of this environment to a VR headset, and rendering a virtual representation of the drone and its environment for the pilot. This way, the pilot can see the immediate environment of the drone up close from a third-person perspective, as opposed to the first-person perspective that most drone cameras provide. This provides much more information about the drone surroundings for the pilot while operating the drone than existing teleoperation solutions. Previous solutions using VR have relied upon pre-made designs of the environment, which makes it difficult to adapt to changing environments. Our solution, in contrast, scans the environment as you fly, making it much more flexible for use in unknown environments. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:14:30 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hughes",
"Jack T.",
""
],
[
"Ghufran",
"Mohammad",
""
],
[
"Rastgoftar",
"Hossein",
""
]
] |
2505.22600 | Kejian Chen | Kejian Chen, Zhengrong Li, Kohei Inayoshi, Luis C. Ho | Dust Budget Crisis in Little Red Dots | 12 pages, 5 figures | null | null | null | astro-ph.GA astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | Little red dots (LRDs), a population of active galactic nuclei (AGNs) recently identified by JWST, are characterized by their compact morphology and red optical continuum emission, which is often interpreted as evidence for significant dust extinction of $A_V \gtrsim 3$ mag. However, the dust-reddened AGN scenario is increasingly challenged by their faint near-to-far infrared emission and a potential "dust budget crisis" in cases when the host galaxy is either undetectably low-mass or absent. In this study, we re-evaluate the dust extinction level in LRDs by modeling the UV-to-infrared spectra for various extinction laws and a broad range of dusty distribution parameters. Comparing the predicted infrared fluxes with observational data from the JWST MIRI, Herschel, and ALMA, our analysis finds that the visual extinction is tightly constrained to $A_V \lesssim 1.0$ mag for A2744-45924 and $A_V \lesssim 1.5$ mag for RUBIES-BLAGN-1 under the SMC extinction laws, with slightly weaker constraints for those with gray extinction in the UV range. The revised $A_V$ values yield a radiative efficiencies of $10\%$ for the LRD population, easing the tension with the Soltan argument for the bulk AGN population at lower redshifts. Moreover, this moderate extinction (or dust-free) scenario, with reprocessed emission spectra testable by future far-infrared observatories, provides a paradigm shift in understanding their natures, environments, and evolutionary pathways of massive black holes in the early universe. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:14:52 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Chen",
"Kejian",
""
],
[
"Li",
"Zhengrong",
""
],
[
"Inayoshi",
"Kohei",
""
],
[
"Ho",
"Luis C.",
""
]
] |
2505.22601 | Jacob Block | Jacob L. Block, Aryan Mokhtari, Sanjay Shakkottai | Machine Unlearning under Overparameterization | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Machine unlearning algorithms aim to remove the influence of specific training samples, ideally recovering the model that would have resulted from training on the remaining data alone. We study unlearning in the overparameterized setting, where many models interpolate the data, and defining the unlearning solution as any loss minimizer over the retained set$\unicode{x2013}$as in prior work in the underparameterized setting$\unicode{x2013}$is inadequate, since the original model may already interpolate the retained data and satisfy this condition. In this regime, loss gradients vanish, rendering prior methods based on gradient perturbations ineffective, motivating both new unlearning definitions and algorithms. For this setting, we define the unlearning solution as the minimum-complexity interpolator over the retained data and propose a new algorithmic framework that only requires access to model gradients on the retained set at the original solution. We minimize a regularized objective over perturbations constrained to be orthogonal to these model gradients, a first-order relaxation of the interpolation condition. For different model classes, we provide exact and approximate unlearning guarantees, and we demonstrate that an implementation of our framework outperforms existing baselines across various unlearning experiments. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:14:57 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Block",
"Jacob L.",
""
],
[
"Mokhtari",
"Aryan",
""
],
[
"Shakkottai",
"Sanjay",
""
]
] |
2505.22602 | Anastasios Kyrillidis | Mahtab Alizadeh Vandchali, Fangshuo (Jasper) Liao, Anastasios Kyrillidis | One Rank at a Time: Cascading Error Dynamics in Sequential Learning | 36 pages | null | null | null | cs.LG cs.AI math.OC | http://creativecommons.org/licenses/by/4.0/ | Sequential learning -- where complex tasks are broken down into simpler, hierarchical components -- has emerged as a paradigm in AI. This paper views sequential learning through the lens of low-rank linear regression, focusing specifically on how errors propagate when learning rank-1 subspaces sequentially. We present an analysis framework that decomposes the learning process into a series of rank-1 estimation problems, where each subsequent estimation depends on the accuracy of previous steps. Our contribution is a characterization of the error propagation in this sequential process, establishing bounds on how errors -- e.g., due to limited computational budgets and finite precision -- affect the overall model accuracy. We prove that these errors compound in predictable ways, with implications for both algorithmic design and stability guarantees. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:16:24 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Vandchali",
"Mahtab Alizadeh",
"",
"Jasper"
],
[
"Fangshuo",
"",
"",
"Jasper"
],
[
"Liao",
"",
""
],
[
"Kyrillidis",
"Anastasios",
""
]
] |
2505.22603 | Jan Hubi\v{c}ka | Dana Barto\v{s}ov\'a, David Chodounsk\'y, Barbara Csima, Jan Hubi\v{c}ka, Mat\v{e}j Kone\v{c}n\'y, Joey Lakerdas-Gayle, Spencer Unger, Andy Zucker | Oscillating subalgebras of the atomless countable Boolean algebra | 9 pages | null | null | null | math.LO cs.DM math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that the big Ramsey degree of the Boolean algebra with 3 atoms within the countable atomless Boolean algebra is infinite. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:17:39 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bartošová",
"Dana",
""
],
[
"Chodounský",
"David",
""
],
[
"Csima",
"Barbara",
""
],
[
"Hubička",
"Jan",
""
],
[
"Konečný",
"Matěj",
""
],
[
"Lakerdas-Gayle",
"Joey",
""
],
[
"Unger",
"Spencer",
""
],
[
"Zucker",
"Andy",
""
]
] |
2505.22604 | Ruixuan Zhang | Ruixuan Zhang, He Wang, Zhengyu Zhao, Zhiqing Guo, Xun Yang, Yunfeng Diao, Meng Wang | Adversarially Robust AI-Generated Image Detection for Free: An Information Theoretic Perspective | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Rapid advances in Artificial Intelligence Generated Images (AIGI) have facilitated malicious use, such as forgery and misinformation. Therefore, numerous methods have been proposed to detect fake images. Although such detectors have been proven to be universally vulnerable to adversarial attacks, defenses in this field are scarce. In this paper, we first identify that adversarial training (AT), widely regarded as the most effective defense, suffers from performance collapse in AIGI detection. Through an information-theoretic lens, we further attribute the cause of collapse to feature entanglement, which disrupts the preservation of feature-label mutual information. Instead, standard detectors show clear feature separation. Motivated by this difference, we propose Training-free Robust Detection via Information-theoretic Measures (TRIM), the first training-free adversarial defense for AIGI detection. TRIM builds on standard detectors and quantifies feature shifts using prediction entropy and KL divergence. Extensive experiments across multiple datasets and attacks validate the superiority of our TRIM, e.g., outperforming the state-of-the-art defense by 33.88% (28.91%) on ProGAN (GenImage), while well maintaining original accuracy. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:20:49 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhang",
"Ruixuan",
""
],
[
"Wang",
"He",
""
],
[
"Zhao",
"Zhengyu",
""
],
[
"Guo",
"Zhiqing",
""
],
[
"Yang",
"Xun",
""
],
[
"Diao",
"Yunfeng",
""
],
[
"Wang",
"Meng",
""
]
] |
2505.22605 | Banafsheh Saber Latibari | Banafsheh Saber Latibari, Najmeh Nazari, Avesta Sasan, Houman Homayoun, Pratik Satam, Soheil Salehi, Hossein Sayadi | Transformers for Secure Hardware Systems: Applications, Challenges, and Outlook | null | null | null | null | cs.CR | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The rise of hardware-level security threats, such as side-channel attacks, hardware Trojans, and firmware vulnerabilities, demands advanced detection mechanisms that are more intelligent and adaptive. Traditional methods often fall short in addressing the complexity and evasiveness of modern attacks, driving increased interest in machine learning-based solutions. Among these, Transformer models, widely recognized for their success in natural language processing and computer vision, have gained traction in the security domain due to their ability to model complex dependencies, offering enhanced capabilities in identifying vulnerabilities, detecting anomalies, and reinforcing system integrity. This survey provides a comprehensive review of recent advancements on the use of Transformers in hardware security, examining their application across key areas such as side-channel analysis, hardware Trojan detection, vulnerability classification, device fingerprinting, and firmware security. Furthermore, we discuss the practical challenges of applying Transformers to secure hardware systems, and highlight opportunities and future research directions that position them as a foundation for next-generation hardware-assisted security. These insights pave the way for deeper integration of AI-driven techniques into hardware security frameworks, enabling more resilient and intelligent defenses. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:22:14 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Latibari",
"Banafsheh Saber",
""
],
[
"Nazari",
"Najmeh",
""
],
[
"Sasan",
"Avesta",
""
],
[
"Homayoun",
"Houman",
""
],
[
"Satam",
"Pratik",
""
],
[
"Salehi",
"Soheil",
""
],
[
"Sayadi",
"Hossein",
""
]
] |
2505.22606 | D. Dominic Briseno-Colunga | D. Dominic Brise\~no-Colunga, Bibek Bhandari, Debmalya Das, Long B. Nguyen, Yosep Kim, David I. Santiago, Irfan Siddiqi, Andrew N. Jordan, and Justin Dressel | Dynamical Sweet and Sour Regions in Bichromatically Driven Floquet Qubits | 16 pages, 5 figures | null | null | null | quant-ph cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | Modern superconducting and semiconducting quantum hardware use external charge and microwave flux drives to both tune and operate devices. However, each external drive is susceptible to low-frequency (e.g., $1/f$) noise that can drastically reduce the decoherence lifetime of the device unless the drive is placed at specific operating points that minimize the sensitivity to fluctuations. We show that operating a qubit in a driven frame using two periodic drives of distinct commensurate frequencies can have advantages over both monochromatically driven frames and static frames with constant offset drives. Employing Floquet theory, we analyze the spectral and lifetime characteristics of a two-level system under weak and strong bichromatic drives, identifying drive-parameter regions with high coherence (sweet spots) and highlighting regions where coherence is limited by additional sensitivity to noise at the drive frequencies (sour spots). We present analytical expressions for quasienergy gaps and dephasing rates, demonstrating that bichromatic driving can alleviate the trade-off between DC and AC noise robustness observed in monochromatic drives. This approach reveals continuous manifolds of doubly dynamical sweet spots, along which drive parameters can be varied without compromising coherence. Our results motivate further study of bichromatic Floquet engineering as a powerful strategy for maintaining tunability in high-coherence quantum systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:22:35 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Briseño-Colunga",
"D. Dominic",
""
],
[
"Bhandari",
"Bibek",
""
],
[
"Das",
"Debmalya",
""
],
[
"Nguyen",
"Long B.",
""
],
[
"Kim",
"Yosep",
""
],
[
"Santiago",
"David I.",
""
],
[
"Siddiqi",
"Irfan",
""
],
[
"Jordan",
"Andrew N.",
""
],
[
"Dressel",
"Justin",
""
]
] |
2505.22607 | Tatsuro Hikawa | Tatsuro Hikawa | Collapse of the $ \mathfrak{sl}_2 $-triple associated to the $ (k, a) $-generalized Fourier transform | 16 pages | null | null | null | math.RT math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Ben Sa\"id--Kobayashi--{\O}rsted introduced a family of $ \mathfrak{sl}_2 $-triples of differential operators $ \mathbb{H}_{k, a} $, $ \mathbb{E}^+_{k, a} $ and $ \mathbb{E}^-_{k, a} $ on $ \mathbb{R}^N \setminus \{0\} $ indexed by a Dunkl parameter $ k $ and a deformation parameter $ a \neq 0 $. In the present paper, we fix $ k = 0 $ and study the behavior as the parameter $ a $ approaches $ 0 $. In this limit, the Lie algebra $ \mathfrak{g}_a = \operatorname{span}_{\mathbb{R}} \{\mathbb{H}_a, \mathbb{E}^+_a, \mathbb{E}^-_a\} \cong \mathfrak{sl}(2, \mathbb{R}) $ degenerates into a three-dimensional commutative Lie algebra $ \mathfrak{g}_0 $, and its spectral properties change. We describe the simultaneous spectral decomposition for $ \mathfrak{g}_0 $, and discuss formulas for operator semigroups with infinitesimal generators in $ \mathfrak{g}_0 $. In particular, we describe the integral kernel of $ \exp(z \lvert x \rvert^2 \Delta) $ as an infinite series, which, in some low-dimensional cases, can be expressed in a closed form using the theta function. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:23:53 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hikawa",
"Tatsuro",
""
]
] |
2505.22608 | Haoning Xu | Haoning Xu, Zhaoqing Li, Youjun Chen, Huimeng Wang, Guinan Li, Mengzhe Geng, Chengxi Deng, Xunying Liu | Effective and Efficient One-pass Compression of Speech Foundation Models Using Sparsity-aware Self-pinching Gates | Submitted to Interspeech 2025 | null | null | null | cs.SD cs.AI eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a novel approach for speech foundation models compression that tightly integrates model pruning and parameter update into a single stage. Highly compact layer-level tied self-pinching gates each containing only a single learnable threshold are jointly trained with uncompressed models and used in fine-grained neuron level pruning. Experiments conducted on the LibriSpeech-100hr corpus suggest that our approach reduces the number of parameters of wav2vec2.0-base and HuBERT-large models by 65% and 60% respectively, while incurring no statistically significant word error rate (WER) increase on the test-clean dataset. Compared to previously published methods on the same task, our approach not only achieves the lowest WER of 7.05% on the test-clean dataset under a comparable model compression ratio of 4.26x, but also operates with at least 25% less model compression time. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:24:21 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Xu",
"Haoning",
""
],
[
"Li",
"Zhaoqing",
""
],
[
"Chen",
"Youjun",
""
],
[
"Wang",
"Huimeng",
""
],
[
"Li",
"Guinan",
""
],
[
"Geng",
"Mengzhe",
""
],
[
"Deng",
"Chengxi",
""
],
[
"Liu",
"Xunying",
""
]
] |
2505.22609 | Sodiq Adewole | Alanna Hazlett, Naomi Ohashi, Timothy Rodriguez, Sodiq Adewole | Chest Disease Detection In X-Ray Images Using Deep Learning Classification Method | null | null | null | null | eess.IV cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | In this work, we investigate the performance across multiple classification models to classify chest X-ray images into four categories of COVID-19, pneumonia, tuberculosis (TB), and normal cases. We leveraged transfer learning techniques with state-of-the-art pre-trained Convolutional Neural Networks (CNNs) models. We fine-tuned these pre-trained architectures on a labeled medical x-ray images. The initial results are promising with high accuracy and strong performance in key classification metrics such as precision, recall, and F1 score. We applied Gradient-weighted Class Activation Mapping (Grad-CAM) for model interpretability to provide visual explanations for classification decisions, improving trust and transparency in clinical applications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:24:33 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hazlett",
"Alanna",
""
],
[
"Ohashi",
"Naomi",
""
],
[
"Rodriguez",
"Timothy",
""
],
[
"Adewole",
"Sodiq",
""
]
] |
2505.22610 | Tobias Schwarz | Tobias Schwarz (1), Tobias Kamm (1), Alexis Engelke (1) ((1) Technical University of Munich) | TPDE: A Fast Adaptable Compiler Back-End Framework | 23 pages, 10 figures | null | null | null | cs.PL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Fast machine code generation is especially important for fast start-up just-in-time compilation, where the compilation time is part of the end-to-end latency. However, widely used compiler frameworks like LLVM do not prioritize fast compilation and require an extra IR translation step increasing latency even further; and rolling a custom code generator is a substantial engineering effort, especially when targeting multiple architectures.
Therefore, in this paper, we present TPDE, a compiler back-end framework that adapts to existing code representations in SSA form. Using an IR-specific adapter providing canonical access to IR data structures and a specification of the IR semantics, the framework performs one analysis pass and then performs the compilation in just a single pass, combining instruction selection, register allocation, and instruction encoding. The generated target instructions are primarily derived code written in high-level language through LLVM's Machine IR, easing portability to different architectures while enabling optimizations during code generation.
To show the generality of our framework, we build a new back-end for LLVM from scratch targeting x86-64 and AArch64. Performance results on SPECint 2017 show that we can compile LLVM-IR 8--24x faster than LLVM -O0 while being on-par in terms of run-time performance. We also demonstrate the benefits of adapting to domain-specific IRs in JIT contexts, particularly WebAssembly and database query compilation, where avoiding the extra IR translation further reduces compilation latency. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:25:53 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Schwarz",
"Tobias",
"",
"Technical University of Munich"
],
[
"Kamm",
"Tobias",
"",
"Technical University of Munich"
],
[
"Engelke",
"Alexis",
"",
"Technical University of Munich"
]
] |
2505.22611 | Piotr Kami\'nski | Piotr Kami\'nski, Artur Tyliszczak, Witold Elsner and Pawe{\l} Niegodajew | Turbulent separation control with a tilted wavy wall: a promising approach for energy savings in aerodynamic systems | null | Energy 324 (2025): 135758 | 10.1016/j.energy.2025.135758 | null | physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | Recently, Kami\'nski et al. [1] demonstrated that a two-dimensional streamwise waviness with carefully selected amplitude and period can be effectively used in postponement of a flow separation at high Reynolds number which is out of reach for other commonly known passive flow control strategies. This paper demonstrates that this approach can be substantially improved by introducing a novel type of surface waviness characterized by tilting the subsequent waves. The research is performed by applying the Large Eddy Simulation (LES) method allowing for a detailed analysis of both instantaneous and time-averaged flow characteristics. It is shown that the tilting eliminates the occurrence of separation zones in waviness troughs, which minimizes the shape drag and maximizes the wall shear stress. In particular, compared to an optimal classical sinusoidal waviness shape the drag coefficient drops 14%. Simultaneously, 70% better performance in separation control is achieved. Moreover, it is also shown that when this issue is a priority, even further improvement can be achieved, though at the expense of greater drag. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:27:40 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Kamiński",
"Piotr",
""
],
[
"Tyliszczak",
"Artur",
""
],
[
"Elsner",
"Witold",
""
],
[
"Niegodajew",
"Paweł",
""
]
] |
2505.22612 | Christian Gang Liu | C. G. Liu, P. Bodorik, and D. Jutla | BPMN to Smart Contract by Business Analyst | null | https://www.icsc-conference.org/2025 | null | null | cs.SE cs.CR | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper addresses the challenge of creating smart contracts for applications represented using Business Process Management and Notation (BPMN) models. In our prior work we presented a methodology that automates the generation of smart contracts from BPMN models. This approach abstracts the BPMN flow control, making it independent of the underlying blockchain infrastructure, with only the BPMN task elements requiring coding. In subsequent research, we enhanced our approach by adding support for nested transactions and enabling a smart contract repair and/or upgrade. To empower Business Analysts (BAs) to generate smart contracts without relying on software developers, we tackled the challenge of generating smart contracts from BPMN models without assistance of a software developer. We exploit the Decision Model and Notation (DMN) standard to represent the decisions and the business logic of the BPMN task elements and amended our methodology for transformation of BPMN models into smart contracts to support also the generation script to represent the business logic represented by the DMN models. To support such transformation, we describe how the BA documents, using the BPMN elements, the flow of information along with the flow of execution. Thus, if the BA is successful in representing the blockchain application requirements using BPMN and DMN models, our methodology and the tool, called TABS, that we developed as a proof of concept, is used to generate the smart contracts directly from those models without developer assistance. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:28:38 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Liu",
"C. G.",
""
],
[
"Bodorik",
"P.",
""
],
[
"Jutla",
"D.",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.