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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2506.02566 | Shiang Hu | Shiang Hu, Xiaolong Huang, Yifan Hu, Xue Xiang, Xiaoliang Sheng, Debin Zhou, Pedro A. Valdes-Sosa | GANORM: Lifespan Normative Modeling of EEG Network Topology based on Multinational Cross-Spectra | null | null | null | null | cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Charting the lifespan evolutionary trajectory of brain function serves as the normative standard for preventing mental disorders during brain development and aging. Although numerous MRI studies have mapped the structural connectome for young cohorts, the EEG-based functional connectome is unknown to characterize human lifespan, limiting its practical applications for the early detection of brain dysfunctions at the community level. This work aimed to undertake normative modeling from the perspective of EEG network topology. Frequency-dependent scalp EEG functional networks were constructed based on EEG cross-spectra aged 5-97 years from 9 countries and network characteristics were quantified. First, GAMLSS were applied to describe the normative curves of the network characteristics in different frequency bands. Subsequently, addressing the limitations of existing regression approaches for whole brain network analysis, this paper proposed an interpretable encoder-decoder framework, Generative Age-dependent brain Network nORmative Model (GANORM). Building upon this framework, we established an age-dependent normative trajectory of the complete brain network for the entire lifespan. Finally, we validated the effectiveness of the norm using EEG datasets from multiple sites. Subsequently, we evaluated the effectiveness of GANORM, and the tested performances of BPNN showed the R^2 was 0.796, the MAE was 0.081, and the RMSE was 0.013. Following established lifespan brain network norm, GANORM also exhibited good results upon verification using healthy and disease data from various sites. The deviation scores from the normative mean for the healthy control group were significantly smaller than those of the disease group. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:51:10 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Hu",
"Shiang",
""
],
[
"Huang",
"Xiaolong",
""
],
[
"Hu",
"Yifan",
""
],
[
"Xiang",
"Xue",
""
],
[
"Sheng",
"Xiaoliang",
""
],
[
"Zhou",
"Debin",
""
],
[
"Valdes-Sosa",
"Pedro A.",
""
]
] |
2506.02567 | Ruixin Xu | Ruixin Xu, Shiqian Cao, Tingting Bo, Yanyu Liu, Wei Zhou | Universal band center model for the HER activity of non-metal site | null | null | null | null | cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, the hydrogen evolution reaction activities of non-metal sites in the transition metal dichalcogenides with the stoichiometry of MX2 are investigated using the first principles calculations. The trained machine learning model demonstrates that the pz band center, bandgap, and period effect are the key factors influencing the HER activity of MX2. Furthermore, it also reveals that the observed non-scaling law between the pz band center and HER activity in the semiconductor-like materials originates from the bandgap-induced downshift of bonding and antibonding states. In addition to the bandgap, the intrinsic p orbital energy level of the non-metal atoms also contributes to the periodic variation of pz band center. Extended calculations indicate that the descriptor is equally applicable to the other catalysts, suggesting its universality in predicting the HER activity of non-metal sites. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:51:53 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Xu",
"Ruixin",
""
],
[
"Cao",
"Shiqian",
""
],
[
"Bo",
"Tingting",
""
],
[
"Liu",
"Yanyu",
""
],
[
"Zhou",
"Wei",
""
]
] |
2506.02568 | Dongzhe Fan | Dongzhe Fan, Yi Fang, Jiajin Liu, Djellel Difallah, Qiaoyu Tan | MLaGA: Multimodal Large Language and Graph Assistant | null | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large Language Models (LLMs) have demonstrated substantial efficacy in advancing graph-structured data analysis. Prevailing LLM-based graph methods excel in adapting LLMs to text-rich graphs, wherein node attributes are text descriptions. However, their applications to multimodal graphs--where nodes are associated with diverse attribute types, such as texts and images--remain underexplored, despite their ubiquity in real-world scenarios. To bridge the gap, we introduce the Multimodal Large Language and Graph Assistant (MLaGA), an innovative model that adeptly extends LLM capabilities to facilitate reasoning over complex graph structures and multimodal attributes. We first design a structure-aware multimodal encoder to align textual and visual attributes within a unified space through a joint graph pre-training objective. Subsequently, we implement a multimodal instruction-tuning approach to seamlessly integrate multimodal features and graph structures into the LLM through lightweight projectors. Extensive experiments across multiple datasets demonstrate the effectiveness of MLaGA compared to leading baseline methods, achieving superior performance in diverse graph learning tasks under both supervised and transfer learning scenarios. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:52:00 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Fan",
"Dongzhe",
""
],
[
"Fang",
"Yi",
""
],
[
"Liu",
"Jiajin",
""
],
[
"Difallah",
"Djellel",
""
],
[
"Tan",
"Qiaoyu",
""
]
] |
2506.02569 | Aina Sumiysohi | Aina Sumiyoshi and Takahiro Yamamoto | Phonon-Induced Current Noise in Single-Walled Carbon Nanotubes across the Ballistic-Diffusive Crossover | 6 pages, 5 fugures | null | null | null | cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | We theoretically elucidate the system length ($L$) dependence of phonon-induced current noise in carbon nanotubes at room temperature over a broad range, encompassing the quantum ballistic and classical diffusive regimes. The power spectral density for the current noise is maximally enhanced when $L$ is comparable to the mean free path $L_0$ of an electron. In the ballistic limit of $L/L_0\ll 1$, the power spectral density increases in proportion to $L$, whereas in the diffusive limit of $L/L_0\gg 1$, it shows a power-law decay $L^{-\alpha}$ with a scaling parameter $\alpha=3.81$. The noise decay for single-walled carbon nanotubes is faster than that previously predicted based on a simple model because of the various electron-phonon scattering processes and the complex energy dependence of the phonon relaxation time. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:52:02 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Sumiyoshi",
"Aina",
""
],
[
"Yamamoto",
"Takahiro",
""
]
] |
2506.02570 | Aleksei Matanin | Aleksei R. Matanin, Nikita S. Smirnov, Anton I. Ivanov, Victor I. Polozov, Daria A. Moskaleva, Elizaveta I. Malevannaya, Margarita V. Androshuk, Maksim I. Teleganov, Yulia A. Agafonova, Denis E. Shirokov, Alexander V. Andriyash and Ilya A. Rodionov | Superconducting integrated random access quantum memory | null | null | null | null | quant-ph cond-mat.mes-hall | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Microwave quantum memory represents a critical component for the development of quantum repeaters and resource-efficient quantum processors. We report the experimental realization of a novel architecture of superconducting random access quantum memory with cycling storage time, achieved through pulsed control of an RF-SQUID coupling element. The device demonstrates a memory cycle time of 1.51 $\mu s$ and achieves 57.5\% fidelity with preservation of the input pulse shape during the first retrieval interval for near-single-photon level excitations, with subsequent exponential decay characterized by a time constant of 11.44 $\mu s$. This performance represents a several-fold improvement over previously reported implementations. Crucially, we establish that while the proposed active coupler realization introduces no measurable fidelity degradation, the primary limitation arises from impedance matching imperfections. These results highlight the potential of proposed architecture for quantum memory applications while identifying specific avenues for near-unity storage fidelity. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:52:13 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Matanin",
"Aleksei R.",
""
],
[
"Smirnov",
"Nikita S.",
""
],
[
"Ivanov",
"Anton I.",
""
],
[
"Polozov",
"Victor I.",
""
],
[
"Moskaleva",
"Daria A.",
""
],
[
"Malevannaya",
"Elizaveta I.",
""
],
[
"Androshuk",
"Margarita V.",
""
],
[
"Teleganov",
"Maksim I.",
""
],
[
"Agafonova",
"Yulia A.",
""
],
[
"Shirokov",
"Denis E.",
""
],
[
"Andriyash",
"Alexander V.",
""
],
[
"Rodionov",
"Ilya A.",
""
]
] |
2506.02571 | Abhishek Vivekanandan | Abhishek Vivekanandan, Christian Hubschneider, J. Marius Z\"ollner | Contrast & Compress: Learning Lightweight Embeddings for Short Trajectories | Submitted for peer review | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | The ability to retrieve semantically and directionally similar short-range trajectories with both accuracy and efficiency is foundational for downstream applications such as motion forecasting and autonomous navigation. However, prevailing approaches often depend on computationally intensive heuristics or latent anchor representations that lack interpretability and controllability. In this work, we propose a novel framework for learning fixed-dimensional embeddings for short trajectories by leveraging a Transformer encoder trained with a contrastive triplet loss that emphasize the importance of discriminative feature spaces for trajectory data. We analyze the influence of Cosine and FFT-based similarity metrics within the contrastive learning paradigm, with a focus on capturing the nuanced directional intent that characterizes short-term maneuvers. Our empirical evaluation on the Argoverse 2 dataset demonstrates that embeddings shaped by Cosine similarity objectives yield superior clustering of trajectories by both semantic and directional attributes, outperforming FFT-based baselines in retrieval tasks. Notably, we show that compact Transformer architectures, even with low-dimensional embeddings (e.g., 16 dimensions, but qualitatively down to 4), achieve a compelling balance between retrieval performance (minADE, minFDE) and computational overhead, aligning with the growing demand for scalable and interpretable motion priors in real-time systems. The resulting embeddings provide a compact, semantically meaningful, and efficient representation of trajectory data, offering a robust alternative to heuristic similarity measures and paving the way for more transparent and controllable motion forecasting pipelines. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:53:04 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Vivekanandan",
"Abhishek",
""
],
[
"Hubschneider",
"Christian",
""
],
[
"Zöllner",
"J. Marius",
""
]
] |
2506.02572 | Juncheng Zhang | Ping Gong, Jiawei Yi, Shengnan Wang, Juncheng Zhang, Zewen Jin, Ouxiang Zhou, Ruibo Liu, Guanbin Xu, Youhui Bai, Bowen Ye, Kun Yuan, Tong Yang, Gong Zhang, Renhai Chen, Feng Wu, Cheng Li | HATA: Trainable and Hardware-Efficient Hash-Aware Top-k Attention for Scalable Large Model Inference | ACL 2025 findings | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large Language Models (LLMs) have emerged as a pivotal research area, yet the attention module remains a critical bottleneck in LLM inference, even with techniques like KVCache to mitigate redundant computations. While various top-$k$ attention mechanisms have been proposed to accelerate LLM inference by exploiting the inherent sparsity of attention, they often struggled to strike a balance between efficiency and accuracy. In this paper, we introduce HATA (Hash-Aware Top-$k$ Attention), a novel approach that systematically integrates low-overhead learning-to-hash techniques into the Top-$k$ attention process. Different from the existing top-k attention methods which are devoted to seeking an absolute estimation of qk score, typically with a great cost, HATA maps queries and keys into binary hash codes, and acquires the relative qk score order with a quite low cost, which is sufficient for realizing top-k attention. Extensive experiments demonstrate that HATA achieves up to 7.2$\times$ speedup compared to vanilla full attention while maintaining model accuracy. In addition, HATA outperforms the state-of-the-art top-$k$ attention methods in both accuracy and efficiency across multiple mainstream LLM models and diverse tasks. HATA is open source at https://github.com/gpzlx1/HATA. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:53:32 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Gong",
"Ping",
""
],
[
"Yi",
"Jiawei",
""
],
[
"Wang",
"Shengnan",
""
],
[
"Zhang",
"Juncheng",
""
],
[
"Jin",
"Zewen",
""
],
[
"Zhou",
"Ouxiang",
""
],
[
"Liu",
"Ruibo",
""
],
[
"Xu",
"Guanbin",
""
],
[
"Bai",
"Youhui",
""
],
[
"Ye",
"Bowen",
""
],
[
"Yuan",
"Kun",
""
],
[
"Yang",
"Tong",
""
],
[
"Zhang",
"Gong",
""
],
[
"Chen",
"Renhai",
""
],
[
"Wu",
"Feng",
""
],
[
"Li",
"Cheng",
""
]
] |
2506.02573 | Muhammad Falensi Azmi | Muhammad Falensi Azmi, Muhammad Dehan Al Kautsar, Alfan Farizki Wicaksono, Fajri Koto | IndoSafety: Culturally Grounded Safety for LLMs in Indonesian Languages | 25 pages | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Although region-specific large language models (LLMs) are increasingly developed, their safety remains underexplored, particularly in culturally diverse settings like Indonesia, where sensitivity to local norms is essential and highly valued by the community. In this work, we present IndoSafety, the first high-quality, human-verified safety evaluation dataset tailored for the Indonesian context, covering five language varieties: formal and colloquial Indonesian, along with three major local languages: Javanese, Sundanese, and Minangkabau. IndoSafety is constructed by extending prior safety frameworks to develop a taxonomy that captures Indonesia's sociocultural context. We find that existing Indonesian-centric LLMs often generate unsafe outputs, particularly in colloquial and local language settings, while fine-tuning on IndoSafety significantly improves safety while preserving task performance. Our work highlights the critical need for culturally grounded safety evaluation and provides a concrete step toward responsible LLM deployment in multilingual settings. Warning: This paper contains example data that may be offensive, harmful, or biased. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:53:55 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Azmi",
"Muhammad Falensi",
""
],
[
"Kautsar",
"Muhammad Dehan Al",
""
],
[
"Wicaksono",
"Alfan Farizki",
""
],
[
"Koto",
"Fajri",
""
]
] |
2506.02574 | Shuai Yuan | Shuai Yuan, Shuang Chen, Tianwu Lin, Jie Wang, Peng Gong | Dynamic mapping from static labels: remote sensing dynamic sample generation with temporal-spectral embedding | null | null | null | null | eess.IV cs.CV cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Accurate remote sensing geographic mapping depends heavily on representative and timely sample data. However, rapid changes in land surface dynamics necessitate frequent updates, quickly rendering previously collected samples obsolete and imposing significant labor demands for continuous manual updates. In this study, we aim to address this problem by dynamic sample generation using existing single-date static labeled samples. We introduce TasGen, a two-stage automated framework to automatically generate dynamic samples, designed to simultaneously model spectral and temporal dependencies in time-series remote sensing imagery via temporal-spectral embedding, capturing land surface changes without additional manual annotations. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:55:16 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Yuan",
"Shuai",
""
],
[
"Chen",
"Shuang",
""
],
[
"Lin",
"Tianwu",
""
],
[
"Wang",
"Jie",
""
],
[
"Gong",
"Peng",
""
]
] |
2506.02575 | Bassano Vacchini | Bassano Vacchini, Andrea Smirne, and Nina Megier | Classical pair of states as optimal pair for quantum distinguishability quantifiers | 13 pages, 1 figure | null | null | null | quant-ph math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | The capability to quantitatively distinguish quantum states is of great importance for a variety of tasks, and has recently played an important role in the study of quantum reduced dynamics and their characterization in terms of memory effects. A crucial property of quantum distinguishability quantifiers considered in the latter framework is the contractivity under the action of completely positive trace-preserving maps. We show that this requirement warrants that the pairs on which these quantifiers attain their maximal value are pairs of orthogonal, and in this sense classical, states. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:55:25 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Vacchini",
"Bassano",
""
],
[
"Smirne",
"Andrea",
""
],
[
"Megier",
"Nina",
""
]
] |
2506.02577 | Wenyan Yang | Wenyan Yang, Joni Pajarinen | Reachability Weighted Offline Goal-conditioned Resampling | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Offline goal-conditioned reinforcement learning (RL) relies on fixed datasets where many potential goals share the same state and action spaces. However, these potential goals are not explicitly represented in the collected trajectories. To learn a generalizable goal-conditioned policy, it is common to sample goals and state-action pairs uniformly using dynamic programming methods such as Q-learning. Uniform sampling, however, requires an intractably large dataset to cover all possible combinations and creates many unreachable state-goal-action pairs that degrade policy performance. Our key insight is that sampling should favor transitions that enable goal achievement. To this end, we propose Reachability Weighted Sampling (RWS). RWS uses a reachability classifier trained via positive-unlabeled (PU) learning on goal-conditioned state-action values. The classifier maps these values to a reachability score, which is then used as a sampling priority. RWS is a plug-and-play module that integrates seamlessly with standard offline RL algorithms. Experiments on six complex simulated robotic manipulation tasks, including those with a robot arm and a dexterous hand, show that RWS significantly improves performance. In one notable case, performance on the HandBlock-Z task improved by nearly 50 percent relative to the baseline. These results indicate the effectiveness of reachability-weighted sampling. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:55:57 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Yang",
"Wenyan",
""
],
[
"Pajarinen",
"Joni",
""
]
] |
2506.02578 | Maria Isabel Alvarez-Casta\~no | Maria Isabel \'Alvarez-Casta\~no, Ye Pu, and Christophe Moser | High Light-Efficiency Holographic Tomographic Volumetric Additive Manufacturing using a MEMS-based Phase-only Light Modulator | 28 pages, 10 figures, 5 supplementary figures | null | null | null | physics.optics | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Light-based 3D printing, which relies on photocurable resins, has shown the capability to produce complex geometries with high resolution and fidelity. Tomographic Volumetric Additive Manufacturing (TVAM) employs a digital micromirror device (DMD) to project high-speed sequences of amplitude light patterns into a rotating resin volume, enabling rapid fabrication of 3D structures through photopolymerization. Typically, the light projection efficiency in such binary amplitude modulator-based systems is below a few percent. Recent advancements introduced phase encoding in TVAM using binary amplitude modulators, improving depth control and boosting light projection efficiency to approximately 10%. This was achieved by implementing the Lee hologram technique to encode phase into binary amplitude patterns. In this work, we present the first 3D printing platform utilizing a phase-only light modulator (PLM), based on an array of micro-electro-mechanical pistons. Compared to amplitude encoding, phase encoding with the PLM yields a 70-fold increase in laser power efficiency. By coupling this efficient light engine with a speckle reduction method in holographic volumetric additive manufacturing (HoloVAM), we experimentally demonstrate printing across different scales from hundreds of micrometers to centimeters using only digital control. The PLM opens up new avenues in volumetric AM for holographic techniques using low-cost single-mode UV laser diodes. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:59:41 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Álvarez-Castaño",
"Maria Isabel",
""
],
[
"Pu",
"Ye",
""
],
[
"Moser",
"Christophe",
""
]
] |
2506.02579 | Lorenzo Oghittu | Lorenzo Oghittu, Premjith Thekkeppatt, Nirav P. Mehta, Seth T. Rittenhouse, Klaasjan van Druten, Florian Schreck, Arghavan Safavi-Naini | Confinement-induced resonances for the creation of quasi-1D ultra cold gases of alkali--alkaline-earth dimers | 11 pages, 4 figures | null | null | null | cond-mat.quant-gas | http://creativecommons.org/licenses/by/4.0/ | We theoretically investigate the role of confinement-induced resonances (CIRs) in low-dimensional ultracold atomic mixtures for the formation of weakly bound dimers. To this end, we examine the scattering properties of a binary atomic mixture confined by a quasi-one-dimensional (quasi-1D) potential. In this regime, the interspecies two-body interaction is modeled as an effective 1D zero-range pseudopotential, with a coupling strength $g_\mathrm{1D}$ derived as a function of the three-dimensional scattering length $a$. This framework enables the study of CIRs in harmonically confined systems, with particular attention to the case of mismatched transverse trapping frequencies for the two atomic species. Finally, we consider the Bose-Fermi mixture of $^{87}$Rb and $^{87}$Sr, and identify values of the experimentally accessible parameters for which CIRs can be exploited to create weakly bound molecules. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:00:31 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Oghittu",
"Lorenzo",
""
],
[
"Thekkeppatt",
"Premjith",
""
],
[
"Mehta",
"Nirav P.",
""
],
[
"Rittenhouse",
"Seth T.",
""
],
[
"van Druten",
"Klaasjan",
""
],
[
"Schreck",
"Florian",
""
],
[
"Safavi-Naini",
"Arghavan",
""
]
] |
2506.02580 | Xuewen Luo | Xuewen Luo, Fengze Yang, Fan Ding, Xiangbo Gao, Shuo Xing, Yang Zhou, Zhengzhong Tu, and Chenxi Liu | V2X-UniPool: Unifying Multimodal Perception and Knowledge Reasoning for Autonomous Driving | null | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Knowledge-driven autonomous driving systems(ADs) offer powerful reasoning capabilities, but face two critical challenges: limited perception due to the short-sightedness of single-vehicle sensors, and hallucination arising from the lack of real-time environmental grounding. To address these issues, this paper introduces V2X-UniPool, a unified framework that integrates multimodal Vehicle-to-Everything (V2X) data into a time-indexed and language-based knowledge pool. By leveraging a dual-query Retrieval-Augmented Generation (RAG) mechanism, which enables retrieval of both static and dynamic knowledge, our system enables ADs to perform accurate, temporally consistent reasoning over both static environment and dynamic traffic context. Experiments on a real-world cooperative driving dataset demonstrate that V2X-UniPool significantly enhances motion planning accuracy and reasoning capability. Remarkably, it enables even zero-shot vehicle-side models to achieve state-of-the-art performance by leveraging V2X-UniPool, while simultaneously reducing transmission cost by over 99.9\% compared to prior V2X methods. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:00:57 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Luo",
"Xuewen",
""
],
[
"Yang",
"Fengze",
""
],
[
"Ding",
"Fan",
""
],
[
"Gao",
"Xiangbo",
""
],
[
"Xing",
"Shuo",
""
],
[
"Zhou",
"Yang",
""
],
[
"Tu",
"Zhengzhong",
""
],
[
"Liu",
"Chenxi",
""
]
] |
2506.02581 | Surajit Majumder | Paritosh Ranjan, Surajit Majumder, Prodip Roy, Bhuban Padhan | Distributedness based scheduling | 7 pages | null | null | null | cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Efficient utilization of computing resources in a Kubernetes cluster is often constrained by the uneven distribution of pods with similar usage patterns. This paper presents a novel scheduling strategy designed to optimize the distributedness of Kubernetes resources based on their usage magnitude and patterns across CPU, memory, network, and storage. By categorizing resource usage into labels such as "cpu high spike" or "memory medium always," and applying these to deployed pods, the system calculates the variance or distributedness factor of similar resource types across cluster nodes. A lower variance indicates a more balanced distribution. The Kubernetes scheduler is enhanced to consider this factor during scheduling decisions, placing new pods on nodes that minimize resource clustering. Furthermore, the approach supports redistribution of existing pods through simulated scheduling to improve balance. This method is adaptable at the cluster, namespace, or application level and is integrated within the standard Kubernetes scheduler, providing a scalable, label-driven mechanism to improve overall resource efficiency in cloud-native environments. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:02:21 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Ranjan",
"Paritosh",
""
],
[
"Majumder",
"Surajit",
""
],
[
"Roy",
"Prodip",
""
],
[
"Padhan",
"Bhuban",
""
]
] |
2506.02582 | Elyes Ahmed | Elyes Ahmed, Brahim Amaziane, Salaheddine Chabab, St\'ephanie Delage Santacreu, Guillaume Galli\'ero, Olav M{\o}yner and Xavier Raynaud | Modeling and Simulation of Coupled Biochemical and Two-Phase Compositional Flow in Underground Hydrogen Storage | null | null | null | null | physics.comp-ph math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | Integrating microbial activity into underground hydrogen storage models is crucial for simulating long-term reservoir behavior. In this work, we present a coupled framework that incorporates bio-geochemical reactions and compositional flow models within the Matlab Reservoir Simulation Toolbox (MRST). Microbial growth and decay are modeled using a double Monod formulation, with populations influenced by hydrogen and carbon dioxide availability. First, a refined Equation of State (EoS) is employed to accurately capture hydrogen dissolution, thereby improving phase behavior and modeling of microbial activity. The model is then discretized using a cell-centered finite-volume method with implicit Euler time discretization. A fully coupled fully implicit strategy is considered. Our implementation builds upon MRST's compositional module by incorporating the S{\o}reide-Whitson EoS, microbial reaction kinetics, and specific effects such as bio-clogging and molecular diffusion. Through a series of 1D, 2D and 3D simulations, we analyze the effects of microbial-induced bio-geochemical transformations on underground hydrogen storage in porous media.These results highlight that accounting for bio-geochemical effects can substantially impact hydrogen loss, purity, and overall storage performance. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:03:44 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Ahmed",
"Elyes",
""
],
[
"Amaziane",
"Brahim",
""
],
[
"Chabab",
"Salaheddine",
""
],
[
"Santacreu",
"Stéphanie Delage",
""
],
[
"Galliéro",
"Guillaume",
""
],
[
"Møyner",
"Olav",
""
],
[
"Raynaud",
"Xavier",
""
]
] |
2506.02583 | Yang Chen | Chi Jiang, Taixia Shi, Dingding Liang, Lei Gao, Chulun Lin, Yang Chen | Advanced microwave photonic waveform editing: enabling the evolution of radar systems into joint radar and spectrum sensing systems | 28 pages, 10 figures | null | null | null | physics.optics | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In response to the urgent demand for the development of future radar application platforms from single radar functionality towards integrated multi-functional systems, we show an advanced microwave photonic waveform editing method that enables the editing of arbitrary radar waveforms, equipping them with the capability to perform spectrum sensing. This, in turn, expands single-function radar systems into joint radar and spectrum sensing systems. We theoretically define and calculate the accumulation function of an arbitrary waveform after passing through a specific dispersive medium, and utilize this accumulation function to further design a corresponding binary sequence for editing the waveform. After editing, the accumulation function of the edited waveform approximates that of a linearly frequency-modulated signal matching the specific dispersive medium. Thus, the edited waveform can be compressed into a narrow pulse after passing through the dispersive medium, realizing the frequency-to-time mapping for achieving frequency measurement or time-frequency analysis. The concept is verified by a simulation and an experiment. Using a dispersion compensating fiber with a total dispersion of -6817 ps/nm, arbitrary waveforms, including a 7-bit Barker phase-coded waveform, a linearly frequency-modulated waveform, a nonlinearly frequency-modulated waveform, and a waveform with an "E" time-frequency diagram, are edited and further used for microwave frequency measurement and time-frequency analysis in an ultra-wide bandwidth of 36.8 GHz. The temporal resolution and frequency resolution are 2 ns and 0.86 GHz, respectively. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:04:03 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Jiang",
"Chi",
""
],
[
"Shi",
"Taixia",
""
],
[
"Liang",
"Dingding",
""
],
[
"Gao",
"Lei",
""
],
[
"Lin",
"Chulun",
""
],
[
"Chen",
"Yang",
""
]
] |
2506.02584 | Christoph Minixhofer | Sarenne Wallbridge, Christoph Minixhofer, Catherine Lai, Peter Bell | Prosodic Structure Beyond Lexical Content: A Study of Self-Supervised Learning | Accepted at INTERSPEECH 2025 | null | null | null | cs.CL cs.AI eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | People exploit the predictability of lexical structures during text comprehension. Though predictable structure is also present in speech, the degree to which prosody, e.g. intonation, tempo, and loudness, contributes to such structure independently of the lexical content is unclear. This study leverages self-supervised learning (SSL) to examine the temporal granularity of structures in the acoustic correlates of prosody. Representations from our proposed Masked Prosody Model can predict perceptual labels dependent on local information, such as word boundaries, but provide the most value for labels involving longer-term structures, like emotion recognition. Probing experiments across various perceptual labels show strong relative gains over untransformed pitch, energy, and voice activity features. Our results reveal the importance of SSL training objective timescale and highlight the value of complex SSL-encoded structures compared to more constrained classical structures. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:04:03 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Wallbridge",
"Sarenne",
""
],
[
"Minixhofer",
"Christoph",
""
],
[
"Lai",
"Catherine",
""
],
[
"Bell",
"Peter",
""
]
] |
2506.02585 | Chunwei Tian | Chunwei Tian, Mingjian Song, Xiaopeng Fan, Xiangtao Zheng, Bob Zhang and David Zhang | A Tree-guided CNN for image super-resolution | This paper has been accepted for publication in IEEE Transactions on Consumer Electronics. 10 pages, 6 figures. Its code can be obtained at https://github.com/hellloxiaotian/TSRNet | null | null | null | eess.IV cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep convolutional neural networks can extract more accurate structural information via deep architectures to obtain good performance in image super-resolution. However, it is not easy to find effect of important layers in a single network architecture to decrease performance of super-resolution. In this paper, we design a tree-guided CNN for image super-resolution (TSRNet). It uses a tree architecture to guide a deep network to enhance effect of key nodes to amplify the relation of hierarchical information for improving the ability of recovering images. To prevent insufficiency of the obtained structural information, cosine transform techniques in the TSRNet are used to extract cross-domain information to improve the performance of image super-resolution. Adaptive Nesterov momentum optimizer (Adan) is applied to optimize parameters to boost effectiveness of training a super-resolution model. Extended experiments can verify superiority of the proposed TSRNet for restoring high-quality images. Its code can be obtained at https://github.com/hellloxiaotian/TSRNet. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:05:11 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Tian",
"Chunwei",
""
],
[
"Song",
"Mingjian",
""
],
[
"Fan",
"Xiaopeng",
""
],
[
"Zheng",
"Xiangtao",
""
],
[
"Zhang",
"Bob",
""
],
[
"Zhang",
"David",
""
]
] |
2506.02586 | Valentina Abramenko | Valentina Abramenko | Time path of turbulence and multi-fractality of magnetic field in the evolution of an active region | 16 pages, 5 figures | Solar Physics (2025) 300:74 | 10.1007/s11207-025-02484-2 | null | astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | Magnetograms acquired with the Helioseismic and Magnetic Imager (HMI) on board the Solar Dynamics Observatory (SDO) were used to calculate and analyze time variations of turbulence and multifractality in the photosphere during the development and flaring of a mature active region NOAA 13354 during its passage across the solar disk. Turbulence was explored with 2D magnetic power spectra from magnetograms, and multifractality was analyzed using the structure functions of magnetograms.
Time variations of the magnetic power spectrum exponent $\alpha$ and of the multifractalty exponent $\kappa$ demonstrate no pre-flare or post-flare abrupt peculiarities, instead, long periods of stability with smooth transitions into other conditions were observed. A conclusion was inferred that the turbulence and multifractality time path in the photospheric magnetic field does not follow the timing of single flares, however, it tends to correspond to the levels of the magneto-morphological complexity and flaring productivity of an AR. So, in the sense of self-organized criticality (SOC), the photosphere, being in the state of self-organization, evolves independently from the highly intermittent, SOC-state corona. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:07:15 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Abramenko",
"Valentina",
""
]
] |
2506.02587 | Weiduo Yuan | Weiduo Yuan, Jerry Li, Justin Yue, Divyank Shah, Konstantinos Karydis, Hang Qiu | BEVCALIB: LiDAR-Camera Calibration via Geometry-Guided Bird's-Eye View Representations | null | null | null | null | cs.CV cs.RO | http://creativecommons.org/licenses/by/4.0/ | Accurate LiDAR-camera calibration is fundamental to fusing multi-modal perception in autonomous driving and robotic systems. Traditional calibration methods require extensive data collection in controlled environments and cannot compensate for the transformation changes during the vehicle/robot movement. In this paper, we propose the first model that uses bird's-eye view (BEV) features to perform LiDAR camera calibration from raw data, termed BEVCALIB. To achieve this, we extract camera BEV features and LiDAR BEV features separately and fuse them into a shared BEV feature space. To fully utilize the geometric information from the BEV feature, we introduce a novel feature selector to filter the most important features in the transformation decoder, which reduces memory consumption and enables efficient training. Extensive evaluations on KITTI, NuScenes, and our own dataset demonstrate that BEVCALIB establishes a new state of the art. Under various noise conditions, BEVCALIB outperforms the best baseline in the literature by an average of (47.08%, 82.32%) on KITTI dataset, and (78.17%, 68.29%) on NuScenes dataset, in terms of (translation, rotation), respectively. In the open-source domain, it improves the best reproducible baseline by one order of magnitude. Our code and demo results are available at https://cisl.ucr.edu/BEVCalib. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:07:18 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Yuan",
"Weiduo",
""
],
[
"Li",
"Jerry",
""
],
[
"Yue",
"Justin",
""
],
[
"Shah",
"Divyank",
""
],
[
"Karydis",
"Konstantinos",
""
],
[
"Qiu",
"Hang",
""
]
] |
2506.02588 | Alessio Zaccone | Wei Chu, Zheng Wang, Christopher Ness, Konrad Samwer, Alessio Zaccone, Lina Hu | Emergent rigidity percolation of five-fold aggregates enables controllable glass properties | null | null | null | null | cond-mat.soft cond-mat.dis-nn cond-mat.mtrl-sci cond-mat.stat-mech physics.chem-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Metallic glasses possess outstanding mechanical and physical properties, making them promising candidates for advanced structural and functional applications; however, the lack of understanding and control over their glass transition and solidification processes remains a significant barrier to practical design. The glass transition from liquid to amorphous solid has remained an open problem in physics despite many theories and recent advances in computational efforts. The question of identifying a clear and well-defined diverging length scale accompanying the glass transition has remained unanswered, as has the nature of the transition and, indeed, the presence of a transition at all, as opposed to a mere dynamical crossover. Here we answer these questions using numerical results and theoretical analysis showing that, in atomic (metallic) glass formers, the glass transition coincides with, and is caused by, a continuous rigidity percolation transition from a liquid-like to a solid-like material. The transition occurs as five-fold symmetric atomic clusters progressively aggregate, forming a system-spanning rigid network that marks the onset of mechanical stability. This percolation-driven rigidity growth is accompanied by a sharp increase in the shear modulus G', indicating the emergence of macroscopic solid-like behavior. Beyond this point, which coincides with the Maxwell isostatic point of the percolating structure, dynamical arrest or "freezing-in" prevents further evolution. The long-sought diverging length scale is thus identified as the percolation-driven growth of rigid five-fold clusters, providing a direct link between local structural motifs and macroscopic mechanical properties at the glass transition. These insights offer practical routes to rationally engineer metallic glasses with targeted mechanical stiffness, hardness, and toughness. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:08:58 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Chu",
"Wei",
""
],
[
"Wang",
"Zheng",
""
],
[
"Ness",
"Christopher",
""
],
[
"Samwer",
"Konrad",
""
],
[
"Zaccone",
"Alessio",
""
],
[
"Hu",
"Lina",
""
]
] |
2506.02589 | Maria Levchenko | Maria Levchenko | Evaluating Named Entity Recognition Models for Russian Cultural News Texts: From BERT to LLM | null | null | null | null | cs.CL cs.AI cs.IR | http://creativecommons.org/licenses/by/4.0/ | This paper addresses the challenge of Named Entity Recognition (NER) for person names within the specialized domain of Russian news texts concerning cultural events. The study utilizes the unique SPbLitGuide dataset, a collection of event announcements from Saint Petersburg spanning 1999 to 2019. A comparative evaluation of diverse NER models is presented, encompassing established transformer-based architectures such as DeepPavlov, RoBERTa, and SpaCy, alongside recent Large Language Models (LLMs) including GPT-3.5, GPT-4, and GPT-4o. Key findings highlight the superior performance of GPT-4o when provided with specific prompting for JSON output, achieving an F1 score of 0.93. Furthermore, GPT-4 demonstrated the highest precision at 0.99. The research contributes to a deeper understanding of current NER model capabilities and limitations when applied to morphologically rich languages like Russian within the cultural heritage domain, offering insights for researchers and practitioners. Follow-up evaluation with GPT-4.1 (April 2025) achieves F1=0.94 for both simple and structured prompts, demonstrating rapid progress across model families and simplified deployment requirements. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:11:16 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Levchenko",
"Maria",
""
]
] |
2506.02590 | Dimitris Koutsianos | Dimitrios Koutsianos and Stavros Zacharopoulos and Yannis Panagakis and Themos Stafylakis | Synthetic Speech Source Tracing using Metric Learning | Submitted to Interspeech 2025 | null | null | null | cs.SD cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper addresses source tracing in synthetic speech-identifying generative systems behind manipulated audio via speaker recognition-inspired pipelines. While prior work focuses on spoofing detection, source tracing lacks robust solutions. We evaluate two approaches: classification-based and metric-learning. We tested our methods on the MLAADv5 benchmark using ResNet and self-supervised learning (SSL) backbones. The results show that ResNet achieves competitive performance with the metric learning approach, matching and even exceeding SSL-based systems. Our work demonstrates ResNet's viability for source tracing while underscoring the need to optimize SSL representations for this task. Our work bridges speaker recognition methodologies with audio forensic challenges, offering new directions for combating synthetic media manipulation. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:12:15 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Koutsianos",
"Dimitrios",
""
],
[
"Zacharopoulos",
"Stavros",
""
],
[
"Panagakis",
"Yannis",
""
],
[
"Stafylakis",
"Themos",
""
]
] |
2506.02591 | Minh Duc Bui | Minh Duc Bui, Kyung Eun Park, Goran Glava\v{s}, Fabian David Schmidt, Katharina von der Wense | On Generalization across Measurement Systems: LLMs Entail More Test-Time Compute for Underrepresented Cultures | Accepted to ACL 2025 Main (Camera-Ready Version) | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Measurement systems (e.g., currencies) differ across cultures, but the conversions between them are well defined so that humans can state facts using any measurement system of their choice. Being available to users from diverse cultural backgrounds, large language models (LLMs) should also be able to provide accurate information irrespective of the measurement system at hand. Using newly compiled datasets we test if this is the case for seven open-source LLMs, addressing three key research questions: (RQ1) What is the default system used by LLMs for each type of measurement? (RQ2) Do LLMs' answers and their accuracy vary across different measurement systems? (RQ3) Can LLMs mitigate potential challenges w.r.t. underrepresented systems via reasoning? Our findings show that LLMs default to the measurement system predominantly used in the data. Additionally, we observe considerable instability and variance in performance across different measurement systems. While this instability can in part be mitigated by employing reasoning methods such as chain-of-thought (CoT), this implies longer responses and thereby significantly increases test-time compute (and inference costs), marginalizing users from cultural backgrounds that use underrepresented measurement systems. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:12:28 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Bui",
"Minh Duc",
""
],
[
"Park",
"Kyung Eun",
""
],
[
"Glavaš",
"Goran",
""
],
[
"Schmidt",
"Fabian David",
""
],
[
"von der Wense",
"Katharina",
""
]
] |
2506.02592 | Zhi-Yuan Chen | Zhi-Yuan Chen, Hao Wang, Xinyu Zhang, Enrui Hu, Yankai Lin | Beyond the Surface: Measuring Self-Preference in LLM Judgments | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent studies show that large language models (LLMs) exhibit self-preference bias when serving as judges, meaning they tend to favor their own responses over those generated by other models. Existing methods typically measure this bias by calculating the difference between the scores a judge model assigns to its own responses and those it assigns to responses from other models. However, this approach conflates self-preference bias with response quality, as higher-quality responses from the judge model may also lead to positive score differences, even in the absence of bias. To address this issue, we introduce gold judgments as proxies for the actual quality of responses and propose the DBG score, which measures self-preference bias as the difference between the scores assigned by the judge model to its own responses and the corresponding gold judgments. Since gold judgments reflect true response quality, the DBG score mitigates the confounding effect of response quality on bias measurement. Using the DBG score, we conduct comprehensive experiments to assess self-preference bias across LLMs of varying versions, sizes, and reasoning abilities. Additionally, we investigate two factors that influence and help alleviate self-preference bias: response text style and the post-training data of judge models. Finally, we explore potential underlying mechanisms of self-preference bias from an attention-based perspective. Our code and data are available at https://github.com/zhiyuanc2001/self-preference. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:12:47 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Chen",
"Zhi-Yuan",
""
],
[
"Wang",
"Hao",
""
],
[
"Zhang",
"Xinyu",
""
],
[
"Hu",
"Enrui",
""
],
[
"Lin",
"Yankai",
""
]
] |
2506.02593 | Arnab Debnath | Arnab Debnath, Gregory J. Stein and Jana Kosecka | A Hybrid Approach to Indoor Social Navigation: Integrating Reactive Local Planning and Proactive Global Planning | Accepted at ICRA 2025 | null | null | null | cs.RO | http://creativecommons.org/licenses/by/4.0/ | We consider the problem of indoor building-scale social navigation, where the robot must reach a point goal as quickly as possible without colliding with humans who are freely moving around. Factors such as varying crowd densities, unpredictable human behavior, and the constraints of indoor spaces add significant complexity to the navigation task, necessitating a more advanced approach. We propose a modular navigation framework that leverages the strengths of both classical methods and deep reinforcement learning (DRL). Our approach employs a global planner to generate waypoints, assigning soft costs around anticipated pedestrian locations, encouraging caution around potential future positions of humans. Simultaneously, the local planner, powered by DRL, follows these waypoints while avoiding collisions. The combination of these planners enables the agent to perform complex maneuvers and effectively navigate crowded and constrained environments while improving reliability. Many existing studies on social navigation are conducted in simplistic or open environments, limiting the ability of trained models to perform well in complex, real-world settings. To advance research in this area, we introduce a new 2D benchmark designed to facilitate development and testing of social navigation strategies in indoor environments. We benchmark our method against traditional and RL-based navigation strategies, demonstrating that our approach outperforms both. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:12:55 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Debnath",
"Arnab",
""
],
[
"Stein",
"Gregory J.",
""
],
[
"Kosecka",
"Jana",
""
]
] |
2506.02594 | Ruibo Duan | Ruibo Duan, Yuxin Liu, Xinyao Dong, Chenglin Fan | EALG: Evolutionary Adversarial Generation of Language Model-Guided Generators for Combinatorial Optimization | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Generating challenging instances is crucial for the evaluation and advancement of combinatorial optimization solvers. In this work, we introduce EALG (Evolutionary Adversarial Generation of Language Model-Guided Generators), a novel framework that automates the co-evolution of optimization problem instances and their corresponding heuristic solvers using large language models (LLMs). EALG leverages a mutation-based adversarial approach that dynamically evolves instance generation procedures to create increasingly difficult problems, while simultaneously synthesizing adaptive heuristic algorithms through interactions with LLMs guided by algorithmic structure. Unlike existing approaches that focus solely on static benchmark creation or manual solver design, EALG provides a seamless pipeline from instance generation to solver synthesis. Experimental results demonstrate that EALG generates significantly harder instances than current benchmarks, and its synthesized solvers generalize effectively across a broad spectrum of combinatorial tasks. This work explores a new paradigm for combinatorial optimization that integrates instance generation with solver design, resulting in state-of-the-art performance. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:13:41 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Duan",
"Ruibo",
""
],
[
"Liu",
"Yuxin",
""
],
[
"Dong",
"Xinyao",
""
],
[
"Fan",
"Chenglin",
""
]
] |
2506.02595 | Edgard Pimentel | Edgard A. Pimentel and Erc\'ilia Sousa | Numerical methods for fully nonlinear degenerate diffusions | null | null | null | null | math.NA cs.NA math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose finite difference methods for degenerate fully nonlinear elliptic equations and prove the convergence of the schemes. Our focus is on the pure equation and a related free boundary problem of transmission type. The cornerstone of our argument is a regularisation procedure. It decouples the degeneracy term from the elliptic operator driving the diffusion process. In the free boundary setting, the absence of degenerate ellipticity entails new, genuine difficulties. To bypass them, we resort to the intrinsic properties of the regularised problem. We present numerical experiments supporting our theoretical results. Our methods are flexible, and our approach can be extended to a broader class of non-variational problems. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:13:41 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Pimentel",
"Edgard A.",
""
],
[
"Sousa",
"Ercília",
""
]
] |
2506.02596 | Fan Gao | Fan Gao, Dongyuan Li, Ding Xia, Fei Mi, Yasheng Wang, Lifeng Shang, Baojun Wang | EssayBench: Evaluating Large Language Models in Multi-Genre Chinese Essay Writing | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Chinese essay writing and its evaluation are critical in educational contexts, yet the capabilities of Large Language Models (LLMs) in this domain remain largely underexplored. Existing benchmarks often rely on coarse-grained text quality metrics, largely overlooking the structural and rhetorical complexities of Chinese essays, particularly across diverse genres. To address this gap, we propose \benchName, a multi-genre benchmark specifically designed for Chinese essay writing across four major genres: Argumentative, Narrative, Descriptive, and Expository. We curate and refine a total of 728 real-world prompts to ensure authenticity and meticulously categorize them into the \textit{Open-Ended} and \textit{Constrained} sets to capture diverse writing scenarios. To reliably evaluate generated essays, we develop a fine-grained, genre-specific scoring framework that hierarchically aggregates scores. We further validate our evaluation protocol through a comprehensive human agreement study. Finally, we benchmark 15 large-sized LLMs, analyzing their strengths and limitations across genres and instruction types. With \benchName, we aim to advance LLM-based Chinese essay evaluation and inspire future research on improving essay generation in educational settings. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:14:46 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Gao",
"Fan",
""
],
[
"Li",
"Dongyuan",
""
],
[
"Xia",
"Ding",
""
],
[
"Mi",
"Fei",
""
],
[
"Wang",
"Yasheng",
""
],
[
"Shang",
"Lifeng",
""
],
[
"Wang",
"Baojun",
""
]
] |
2506.02597 | Zheng-Xue Ren | Zhengxue Ren, Serdar Elhatisari, Ulf-G. Mei{\ss}ner | Ab initio study of the radii of oxygen isotopes | 6+6 pages, 2+5 figures | null | null | null | nucl-th hep-lat nucl-ex | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present an {\em ab initio} study of the charge and matter radii of oxygen isotopes from $^{16}$O to $^{20}$O using nuclear lattice effective field theory (NLEFT) with high-fidelity N$^3$LO chiral interactions. To efficiently address the Monte Carlo sign problem encountered in nuclear radius calculations, we introduce the {\em partial pinhole algorithm}, significantly reducing statistical uncertainties and extending the reach to more neutron-rich and proton-rich isotopes. Our computed charge radii for $^{16}$O, $^{17}$O, and $^{18}$O closely match experimental data, and we predict a charge radius of $2.810(32)$ fm for $^{20}$O. The calculated matter radii show excellent agreement with values extracted from low-energy proton and electron elastic scattering data, but are inconsistent with those derived from interaction cross sections and charge-changing cross section measurements. These discrepancies highlight model-dependent ambiguities in the experimental extraction methods of matter radii and underscore the value of precise theoretical benchmarks from NLEFT calculations. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:20:01 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Ren",
"Zhengxue",
""
],
[
"Elhatisari",
"Serdar",
""
],
[
"Meißner",
"Ulf-G.",
""
]
] |
2506.02598 | Marius Nielsen | Christoph Winges, Marius Nielsen | The presentable stable envelope of an exact category | 26 pages, comments are welcome | null | null | null | math.AT math.CT math.KT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We prove an analogue of the Gabriel--Quillen embedding theorem for exact $\infty$-categories, giving rise to a presentable version of Klemenc's stable envelope of an exact $\infty$-category. Moreover, we construct a symmetric monoidal structure on the $\infty$-category of small exact $\infty$-categories and discuss the multiplicative properties of the Gabriel--Quillen embedding. For $E$ an Adams-type homotopy associative ring spectrum, this allows us to identify the symmetric monoidal $\infty$-category of $E$-based synthetic spectra with the presentable stable envelope of the exact $\infty$-category of compact spectra with finite projective $E$-homology.
In addition, we show that algebraic K-theory, considered as a functor on exact $\infty$-categories, admits a unique delooping as a localising invariant. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:21:03 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Winges",
"Christoph",
""
],
[
"Nielsen",
"Marius",
""
]
] |
2506.02599 | Niklas Rossberg | Niklas Ro{\ss}berg, Marion Neumeier, Sinan Hasirlioglu, Mohamed Essayed Bouzouraa and Michael Botsch | Assessing the Completeness of Traffic Scenario Categories for Automated Highway Driving Functions via Cluster-based Analysis | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The ability to operate safely in increasingly complex traffic scenarios is a fundamental requirement for Automated Driving Systems (ADS). Ensuring the safe release of ADS functions necessitates a precise understanding of the occurring traffic scenarios. To support this objective, this work introduces a pipeline for traffic scenario clustering and the analysis of scenario category completeness. The Clustering Vector Quantized - Variational Autoencoder (CVQ-VAE) is employed for the clustering of highway traffic scenarios and utilized to create various catalogs with differing numbers of traffic scenario categories. Subsequently, the impact of the number of categories on the completeness considerations of the traffic scenario categories is analyzed. The results show an outperforming clustering performance compared to previous work. The trade-off between cluster quality and the amount of required data to maintain completeness is discussed based on the publicly available highD dataset. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:24:14 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Roßberg",
"Niklas",
""
],
[
"Neumeier",
"Marion",
""
],
[
"Hasirlioglu",
"Sinan",
""
],
[
"Bouzouraa",
"Mohamed Essayed",
""
],
[
"Botsch",
"Michael",
""
]
] |
2506.02600 | Lucas Lagarde | Lucas Lagarde | Unramified Brauer groups of homogeneous spaces with finite stabilisers and the Grunwald Problem | 41 pages, comments welcome | null | null | null | math.AG math.NT | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We provide an algorithm for calculating the unramified Brauer group of a homogeneous space $X$ of a semi-simple simply connected group $H$ with finite geometric stabiliser over any field of characteristic 0. When $k$ is a number field, we use the obtained description of the unramified Brauer group in order to study the Brauer-Manin obstruction to weak approximation on $X$. In particular, we provide an algorithm to compute the Brauer-Manin obstruction on $X$, which guarantees effectivity of the Grunwald problem for supersolvable groups thanks to previous work of Harpaz and Wittenberg. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:24:36 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Lagarde",
"Lucas",
""
]
] |
2506.02601 | Shiyu Shen | Shiyu Shen, Bin Pan, Ziye Zhang and Zhenwei Shi | Hyperspectral Image Generation with Unmixing Guided Diffusion Model | null | null | null | null | cs.CV eess.IV | http://creativecommons.org/licenses/by/4.0/ | Recently, hyperspectral image generation has received increasing attention, but existing generative models rely on conditional generation schemes, which limits the diversity of generated images. Diffusion models are popular for their ability to generate high-quality samples, but adapting these models from RGB to hyperspectral data presents the challenge of high dimensionality and physical constraints. To address these challenges, we propose a novel diffusion model guided by hyperspectral unmixing. Our model comprises two key modules: an unmixing autoencoder module and an abundance diffusion module. The unmixing autoencoder module leverages unmixing guidance to shift the generative task from the image space to the low-dimensional abundance space, significantly reducing computational complexity while preserving high fidelity. The abundance diffusion module generates samples that satisfy the constraints of non-negativity and unity, ensuring the physical consistency of the reconstructed HSIs. Additionally, we introduce two evaluation metrics tailored to hyperspectral data. Empirical results, evaluated using both traditional metrics and our proposed metrics, indicate that our model is capable of generating high-quality and diverse hyperspectral images, offering an advancement in hyperspectral data generation. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:27:10 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Shen",
"Shiyu",
""
],
[
"Pan",
"Bin",
""
],
[
"Zhang",
"Ziye",
""
],
[
"Shi",
"Zhenwei",
""
]
] |
2506.02602 | Wang Yuqi | Yuqi Wang, Yadong Zeng, Ralf Deiterding, Jianhan Liang | An efficient GPU-accelerated adaptive mesh refinement framework for high-fidelity compressible reactive flows modeling | null | null | null | null | physics.comp-ph | http://creativecommons.org/licenses/by-nc-sa/4.0/ | This paper presents a heterogeneous adaptive mesh refinement (AMR) framework for efficient simulation of moderately stiff reactive problems. This framework features an elaborate subcycling-in-time algorithm along with a specialized refluxing algorithm, all unified in a highly parallel codebase. We have also developed a low-storage variant of explicit chemical integrators by optimizing the register usage of GPU, achieving respectively 6x and 3x times speedups as compared to the implicit and standard explicit methods with comparable order of accuracy. A suite of benchmarks have confirmed the framework's fidelity for both non-reactive and reactive simulations with/without AMR. By leveraging our parallelization strategy that is developed on AMReX, we have demonstrated remarkable speedups on various problems on a NVIDIA V100 GPU than using a Intel i9 CPU within the same codebase; in problems with complex physics and spatiotemporally distributed stiffness such as the hydrogen detonation propagation, we have achieved an overall 6.49x acceleration ratio. The computation scalability of the framework is also validated through the weak scaling test, demonstrating excellent parallel efficiency across multiple GPU nodes. At last, a practical application of this GPU-accelerated SAMR framework to large-scale direct numerical simulations is demonstrated by successful simulation of the three-dimensional reactive shock-bubble interaction problem; we have saved significant computational costs while maintaining the comparable accuracy, as compared to a prior uniform DNS study performed on CPUs. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:27:22 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Wang",
"Yuqi",
""
],
[
"Zeng",
"Yadong",
""
],
[
"Deiterding",
"Ralf",
""
],
[
"Liang",
"Jianhan",
""
]
] |
2506.02603 | Jose Manuel Camacho Rodriguez | Jose Manuel Camacho, Roi Naveiro, David Rios Insua | Computational adversarial risk analysis for general security games | 35 pages, 6 tables, 17 figures | null | null | null | cs.GT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper provides an efficient computational scheme to handle general security games from an adversarial risk analysis perspective. Two cases in relation to single-stage and multi-stage simultaneous defend-attack games motivate our approach to general setups which uses bi-agent influence diagrams as underlying problem structure and augmented probability simulation as core computational methodology. Theoretical convergence and numerical, modeling, and implementation issues are thoroughly discussed. A disinformation war case study illustrates the relevance of the proposed approach. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:28:06 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Camacho",
"Jose Manuel",
""
],
[
"Naveiro",
"Roi",
""
],
[
"Insua",
"David Rios",
""
]
] |
2506.02604 | Chunwei Tian | Tian Chunwei, Song Mingjian, Zuo Wangmeng, Du Bo, Zhang Yanning and Zhang Shichao | Application of convolutional neural networks in image super-resolution | It has been accepted by CAAI transactions on intelligent systems, in Chinese language | null | null | null | cs.CV eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Due to strong learning abilities of convolutional neural networks (CNNs), they have become mainstream methods for image super-resolution. However, there are big differences of different deep learning methods with different types. There is little literature to summarize relations and differences of different methods in image super-resolution. Thus, summarizing these literatures are important, according to loading capacity and execution speed of devices. This paper first introduces principles of CNNs in image super-resolution, then introduces CNNs based bicubic interpolation, nearest neighbor interpolation, bilinear interpolation, transposed convolution, sub-pixel layer, meta up-sampling for image super-resolution to analyze differences and relations of different CNNs based interpolations and modules, and compare performance of these methods by experiments. Finally, this paper gives potential research points and drawbacks and summarizes the whole paper, which can facilitate developments of CNNs in image super-resolution. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:28:08 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Chunwei",
"Tian",
""
],
[
"Mingjian",
"Song",
""
],
[
"Wangmeng",
"Zuo",
""
],
[
"Bo",
"Du",
""
],
[
"Yanning",
"Zhang",
""
],
[
"Shichao",
"Zhang",
""
]
] |
2506.02605 | Xue Wu | Xue Wu, Jingwei Xin, Zhijun Tu, Jie Hu, Jie Li, Nannan Wang, and Xinbo Gao | One-Step Diffusion-based Real-World Image Super-Resolution with Visual Perception Distillation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Diffusion-based models have been widely used in various visual generation tasks, showing promising results in image super-resolution (SR), while typically being limited by dozens or even hundreds of sampling steps. Although existing methods aim to accelerate the inference speed of multi-step diffusion-based SR methods through knowledge distillation, their generated images exhibit insufficient semantic alignment with real images, resulting in suboptimal perceptual quality reconstruction, specifically reflected in the CLIPIQA score. These methods still have many challenges in perceptual quality and semantic fidelity. Based on the challenges, we propose VPD-SR, a novel visual perception diffusion distillation framework specifically designed for SR, aiming to construct an effective and efficient one-step SR model. Specifically, VPD-SR consists of two components: Explicit Semantic-aware Supervision (ESS) and High-Frequency Perception (HFP) loss. Firstly, the ESS leverages the powerful visual perceptual understanding capabilities of the CLIP model to extract explicit semantic supervision, thereby enhancing semantic consistency. Then, Considering that high-frequency information contributes to the visual perception quality of images, in addition to the vanilla distillation loss, the HFP loss guides the student model to restore the missing high-frequency details in degraded images that are critical for enhancing perceptual quality. Lastly, we expand VPD-SR in adversarial training manner to further enhance the authenticity of the generated content. Extensive experiments conducted on synthetic and real-world datasets demonstrate that the proposed VPD-SR achieves superior performance compared to both previous state-of-the-art methods and the teacher model with just one-step sampling. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:28:13 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Wu",
"Xue",
""
],
[
"Xin",
"Jingwei",
""
],
[
"Tu",
"Zhijun",
""
],
[
"Hu",
"Jie",
""
],
[
"Li",
"Jie",
""
],
[
"Wang",
"Nannan",
""
],
[
"Gao",
"Xinbo",
""
]
] |
2506.02607 | Matthias Irlbeck | Matthias Irlbeck, Zakhar Kabluchko, Tobias M\"uller | On the shape of the typical Poisson-Voronoi cell in high dimensions | null | null | null | null | math.PR | http://creativecommons.org/licenses/by/4.0/ | We study the typical cell of the Poisson-Voronoi tessellation. We show that when divided by the $d$-th root of the intensity parameter $\lambda$ of the Poisson process times the volume of the unit ball, the inradius, outradius, diameter and mean width of the typical cell converge in probability to the constants $1/2, 1, 2, 2$ respectively, as the dimension $d\to\infty$. We also show that the width of the typical cell, when rescaled in the same way, is bounded between $2\sqrt{5}/(2+\sqrt{5})-o_d(1)$ and $3/2+o_d(1)$, with probability $1-o_d(1)$. These results in particular imply that, with probability $1-o_d(1)$, the Hausdorff distance between the typical cell and any ball is at least of the order of the diameter of the typical cell.
In addition, we show that for all $k$ with $d-k\to\infty$, with probability $1-o_d(1)$, all faces of dimension $k$ have a diameter that is of a much smaller order than the diameter, inradius, etc., of the full typical cell. The same is true for ''almost all'' faces of dimension $d-k$ with $k$ fixed. And, we show that the number of such faces is $\left( (k+1)^{(k+1)/2} / k^{k/2} \pm o_d(1) \right)^d$ with probability $1-o_d(1)$. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:28:26 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Irlbeck",
"Matthias",
""
],
[
"Kabluchko",
"Zakhar",
""
],
[
"Müller",
"Tobias",
""
]
] |
2506.02608 | Luca Fabbri | Luca Fabbri, Stefano Vignolo, Giuseppe De Maria, Sante Carloni | Dirac Fields in Hydrodynamic Form and their Thermodynamic Formulation | 13 pages | null | null | null | math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider the theory of spinor fields written in polar form and we re-express it in terms of the so-called $1\!+\!1\!+\!2$ covariant splitting: after this is done for the basic kinematic variables, we proceed to decompose the dynamical equations, both for the case of the Dirac differential field equations and for the case of the energy density tensor. As an explicit example of a real physical application we deal with the hydrogen atom. Comments are addressed in the end. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:28:36 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Fabbri",
"Luca",
""
],
[
"Vignolo",
"Stefano",
""
],
[
"De Maria",
"Giuseppe",
""
],
[
"Carloni",
"Sante",
""
]
] |
2506.02609 | Tianfan Jiang | Tianfan Jiang and Mei Wu and Wenchao Weng and Dewen Seng and Yiqian Lin | A Time-Enhanced Data Disentanglement Network for Traffic Flow Forecasting | null | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In recent years, traffic flow prediction has become a highlight in the field of intelligent transportation systems. However, due to the temporal variations and dynamic spatial correlations of traffic data, traffic prediction remains highly challenging.Traditional spatiotemporal networks, which rely on end-to-end training, often struggle to handle the diverse data dependencies of multiple traffic flow patterns. Additionally, traffic flow variations are highly sensitive to temporal information changes. Regrettably, other researchers have not sufficiently recognized the importance of temporal information.To address these challenges, we propose a novel approach called A Time-Enhanced Data Disentanglement Network for Traffic Flow Forecasting (TEDDN). This network disentangles the originally complex and intertwined traffic data into stable patterns and trends. By flexibly learning temporal and node information through a dynamic graph enhanced by a temporal feature extraction module, TEDDN demonstrates significant efficacy in disentangling and extracting complex traffic information. Experimental evaluations and ablation studies on four real-world datasets validate the superiority of our method. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:28:48 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Jiang",
"Tianfan",
""
],
[
"Wu",
"Mei",
""
],
[
"Weng",
"Wenchao",
""
],
[
"Seng",
"Dewen",
""
],
[
"Lin",
"Yiqian",
""
]
] |
2506.02610 | Zhaoyang Li | Zhaoyang Li, Jie Wang, XiaoXiao Li, Wangjie Li, Longjie Luo, Lin Li, Qingyang Hong | Speaker Diarization with Overlapping Community Detection Using Graph Attention Networks and Label Propagation Algorithm | null | null | null | null | cs.SD cs.AI eess.AS | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In speaker diarization, traditional clustering-based methods remain widely used in real-world applications. However, these methods struggle with the complex distribution of speaker embeddings and overlapping speech segments. To address these limitations, we propose an Overlapping Community Detection method based on Graph Attention networks and the Label Propagation Algorithm (OCDGALP). The proposed framework comprises two key components: (1) a graph attention network that refines speaker embeddings and node connections by aggregating information from neighboring nodes, and (2) a label propagation algorithm that assigns multiple community labels to each node, enabling simultaneous clustering and overlapping community detection. Experimental results show that the proposed method significantly reduces the Diarization Error Rate (DER), achieving a state-of-the-art 15.94% DER on the DIHARD-III dataset without oracle Voice Activity Detection (VAD), and an impressive 11.07% with oracle VAD. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:29:10 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Li",
"Zhaoyang",
""
],
[
"Wang",
"Jie",
""
],
[
"Li",
"XiaoXiao",
""
],
[
"Li",
"Wangjie",
""
],
[
"Luo",
"Longjie",
""
],
[
"Li",
"Lin",
""
],
[
"Hong",
"Qingyang",
""
]
] |
2506.02611 | Tanguy Lions | Timothy Budd and Tanguy Lions | The tight length spectrum of large-genus random hyperbolic surfaces with many cusps | 43 pages, 9 figures | null | null | null | math.PR math.GT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Since the work of Mirzakhani \& Petri \cite{Mirzakhani_petri_2019} on random hyperbolic surfaces of large genus, length statistics of closed geodesics have been studied extensively. We focus on the case of random hyperbolic surfaces with cusps, the number $n_g$ of which grows with the genus $g$. We prove that if $n_g$ grows fast enough and we restrict attention to special geodesics that are \emph{tight}, we recover upon proper normalization the same Poisson point process in the large-$g$ limit for the length statistics. The proof relies on a recursion formula for tight Weil-Petersson volumes obtained in \cite{budd2023topological} and on a generalization of Mirzakhani's integration formula to the tight setting. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:29:24 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Budd",
"Timothy",
""
],
[
"Lions",
"Tanguy",
""
]
] |
2506.02612 | Jan Robine | Jan Robine, Marc H\"oftmann, Stefan Harmeling | Simple, Good, Fast: Self-Supervised World Models Free of Baggage | Published as a conference paper at ICLR 2025. Code is available at https://github.com/jrobine/sgf | null | null | null | cs.LG cs.AI stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | What are the essential components of world models? How far do we get with world models that are not employing RNNs, transformers, discrete representations, and image reconstructions? This paper introduces SGF, a Simple, Good, and Fast world model that uses self-supervised representation learning, captures short-time dependencies through frame and action stacking, and enhances robustness against model errors through data augmentation. We extensively discuss SGF's connections to established world models, evaluate the building blocks in ablation studies, and demonstrate good performance through quantitative comparisons on the Atari 100k benchmark. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:29:32 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Robine",
"Jan",
""
],
[
"Höftmann",
"Marc",
""
],
[
"Harmeling",
"Stefan",
""
]
] |
2506.02613 | Weihai Zhang | Xiushan Jiang, Weihai Zhang | Learning-based primal-dual optimal control of discrete-time stochastic systems with multiplicative noise | null | null | null | null | math.OC | http://creativecommons.org/licenses/by/4.0/ | Reinforcement learning (RL) is an effective approach for solving optimal control problems without knowing the exact information of the system model. However, the classical Q-learning method, a model-free RL algorithm, has its limitations, such as lack of strict theoretical analysis and the need for artificial disturbances during implementation. This paper explores the partially model-free stochastic linear quadratic regulator (SLQR) problem for a system with multiplicative noise from the primal-dual perspective to address these challenges. This approach lays a strong theoretical foundation for understanding the intrinsic mechanisms of classical RL algorithms. We reformulate the SLQR into a non-convex primal-dual optimization problem and derive a strong duality result, which enables us to provide model-based and model-free algorithms for SLQR optimal policy design based on the Karush-Kuhn-Tucker (KKT) conditions. An illustrative example demonstrates the proposed model-free algorithm's validity, showcasing the central nervous system's learning mechanism in human arm movement. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:30:13 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Jiang",
"Xiushan",
""
],
[
"Zhang",
"Weihai",
""
]
] |
2506.02615 | Safaa Abdullahi Moallim Mohamud | Safaa Abdullahi Moallim Mohamud, Minjin Baek, and Dong Seog Han | Hierarchical Question-Answering for Driving Scene Understanding Using Vision-Language Models | This work has been submitted to the IEEE for possible publication | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we present a hierarchical question-answering (QA) approach for scene understanding in autonomous vehicles, balancing cost-efficiency with detailed visual interpretation. The method fine-tunes a compact vision-language model (VLM) on a custom dataset specific to the geographical area in which the vehicle operates to capture key driving-related visual elements. At the inference stage, the hierarchical QA strategy decomposes the scene understanding task into high-level and detailed sub-questions. Instead of generating lengthy descriptions, the VLM navigates a structured question tree, where answering high-level questions (e.g., "Is it possible for the ego vehicle to turn left at the intersection?") triggers more detailed sub-questions (e.g., "Is there a vehicle approaching the intersection from the opposite direction?"). To optimize inference time, questions are dynamically skipped based on previous answers, minimizing computational overhead. The extracted answers are then synthesized using handcrafted templates to ensure coherent, contextually accurate scene descriptions. We evaluate the proposed approach on the custom dataset using GPT reference-free scoring, demonstrating its competitiveness with state-of-the-art methods like GPT-4o in capturing key scene details while achieving significantly lower inference time. Moreover, qualitative results from real-time deployment highlight the proposed approach's capacity to capture key driving elements with minimal latency. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:32:43 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Mohamud",
"Safaa Abdullahi Moallim",
""
],
[
"Baek",
"Minjin",
""
],
[
"Han",
"Dong Seog",
""
]
] |
2506.02616 | Qi Liao | Qi Liao, Parijat Bhattacharjee | Compositional Learning for Modular Multi-Agent Self-Organizing Networks | null | IEEE ICMLCN 2025 | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Self-organizing networks face challenges from complex parameter interdependencies and conflicting objectives. This study introduces two compositional learning approaches-Compositional Deep Reinforcement Learning (CDRL) and Compositional Predictive Decision-Making (CPDM)-and evaluates their performance under training time and safety constraints in multi-agent systems. We propose a modular, two-tier framework with cell-level and cell-pair-level agents to manage heterogeneous agent granularities while reducing model complexity. Numerical simulations reveal a significant reduction in handover failures, along with improved throughput and latency, outperforming conventional multi-agent deep reinforcement learning approaches. The approach also demonstrates superior scalability, faster convergence, higher sample efficiency, and safer training in large-scale self-organizing networks. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:33:18 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Liao",
"Qi",
""
],
[
"Bhattacharjee",
"Parijat",
""
]
] |
2506.02617 | Xinyi Hou | Yinglin Xie, Xinyi Hou, Yanjie Zhao, Shenao Wang, Kai Chen and Haoyu Wang | Toward Understanding Bugs in Vector Database Management Systems | null | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Vector database management systems (VDBMSs) play a crucial role in facilitating semantic similarity searches over high-dimensional embeddings from diverse data sources. While VDBMSs are widely used in applications such as recommendation, retrieval-augmented generation (RAG), and multimodal search, their reliability remains underexplored. Traditional database reliability models cannot be directly applied to VDBMSs because of fundamental differences in data representation, query mechanisms, and system architecture. To address this gap, we present the first large-scale empirical study of software defects in VDBMSs. We manually analyzed 1,671 bug-fix pull requests from 15 widely used open-source VDBMSs and developed a comprehensive taxonomy of bugs based on symptoms, root causes, and developer fix strategies. Our study identifies five categories of bug symptoms, with more than half manifesting as functional failures. We further reveal 31 recurring fault patterns and highlight failure modes unique to vector search systems. In addition, we summarize 12 common fix strategies, whose distribution underscores the critical importance of correct program logic. These findings provide actionable insights into VDBMS reliability challenges and offer guidance for building more robust future systems. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:34:01 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Xie",
"Yinglin",
""
],
[
"Hou",
"Xinyi",
""
],
[
"Zhao",
"Yanjie",
""
],
[
"Wang",
"Shenao",
""
],
[
"Chen",
"Kai",
""
],
[
"Wang",
"Haoyu",
""
]
] |
2506.02618 | Jialiang Zhang | Jialiang Zhang, Haoran Geng, Yang You, Congyue Deng, Pieter Abbeel, Jitendra Malik, Leonidas Guibas | Rodrigues Network for Learning Robot Actions | null | null | null | null | cs.RO cs.CV | http://creativecommons.org/licenses/by/4.0/ | Understanding and predicting articulated actions is important in robot learning. However, common architectures such as MLPs and Transformers lack inductive biases that reflect the underlying kinematic structure of articulated systems. To this end, we propose the Neural Rodrigues Operator, a learnable generalization of the classical forward kinematics operation, designed to inject kinematics-aware inductive bias into neural computation. Building on this operator, we design the Rodrigues Network (RodriNet), a novel neural architecture specialized for processing actions. We evaluate the expressivity of our network on two synthetic tasks on kinematic and motion prediction, showing significant improvements compared to standard backbones. We further demonstrate its effectiveness in two realistic applications: (i) imitation learning on robotic benchmarks with the Diffusion Policy, and (ii) single-image 3D hand reconstruction. Our results suggest that integrating structured kinematic priors into the network architecture improves action learning in various domains. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:34:06 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Zhang",
"Jialiang",
""
],
[
"Geng",
"Haoran",
""
],
[
"You",
"Yang",
""
],
[
"Deng",
"Congyue",
""
],
[
"Abbeel",
"Pieter",
""
],
[
"Malik",
"Jitendra",
""
],
[
"Guibas",
"Leonidas",
""
]
] |
2506.02619 | Chongxu Wang | Yanbei Liu, Chongxu Wang, Zhitao Xiao, Lei Geng, Yanwei Pang, Xiao Wang | HGOT: Self-supervised Heterogeneous Graph Neural Network with Optimal Transport | The paper has 9 pages of text and 13 pages in total (including acknowledgments, impact statement, references, and appendix), with 6 figures and 2 tables. This paper has been accepted by ICML 2025 conference and this is a final version of the manuscript submitted to the conference | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Heterogeneous Graph Neural Networks (HGNNs), have demonstrated excellent capabilities in processing heterogeneous information networks. Self-supervised learning on heterogeneous graphs, especially contrastive self-supervised strategy, shows great potential when there are no labels. However, this approach requires the use of carefully designed graph augmentation strategies and the selection of positive and negative samples. Determining the exact level of similarity between sample pairs is non-trivial.To solve this problem, we propose a novel self-supervised Heterogeneous graph neural network with Optimal Transport (HGOT) method which is designed to facilitate self-supervised learning for heterogeneous graphs without graph augmentation strategies. Different from traditional contrastive self-supervised learning, HGOT employs the optimal transport mechanism to relieve the laborious sampling process of positive and negative samples. Specifically, we design an aggregating view (central view) to integrate the semantic information contained in the views represented by different meta-paths (branch views). Then, we introduce an optimal transport plan to identify the transport relationship between the semantics contained in the branch view and the central view. This allows the optimal transport plan between graphs to align with the representations, forcing the encoder to learn node representations that are more similar to the graph space and of higher quality. Extensive experiments on four real-world datasets demonstrate that our proposed HGOT model can achieve state-of-the-art performance on various downstream tasks. In particular, in the node classification task, HGOT achieves an average of more than 6% improvement in accuracy compared with state-of-the-art methods. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:35:29 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Liu",
"Yanbei",
""
],
[
"Wang",
"Chongxu",
""
],
[
"Xiao",
"Zhitao",
""
],
[
"Geng",
"Lei",
""
],
[
"Pang",
"Yanwei",
""
],
[
"Wang",
"Xiao",
""
]
] |
2506.02620 | Dongyu Yan | Dongyu Yan, Leyi Wu, Jiantao Lin, Luozhou Wang, Tianshuo Xu, Zhifei Chen, Zhen Yang, Lie Xu, Shunsi Zhang, Yingcong Chen | FlexPainter: Flexible and Multi-View Consistent Texture Generation | 11 pages, 10 figures in main paper, 10 pages, 12 figures in supplementary | null | null | null | cs.GR cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Texture map production is an important part of 3D modeling and determines the rendering quality. Recently, diffusion-based methods have opened a new way for texture generation. However, restricted control flexibility and limited prompt modalities may prevent creators from producing desired results. Furthermore, inconsistencies between generated multi-view images often lead to poor texture generation quality. To address these issues, we introduce \textbf{FlexPainter}, a novel texture generation pipeline that enables flexible multi-modal conditional guidance and achieves highly consistent texture generation. A shared conditional embedding space is constructed to perform flexible aggregation between different input modalities. Utilizing such embedding space, we present an image-based CFG method to decompose structural and style information, achieving reference image-based stylization. Leveraging the 3D knowledge within the image diffusion prior, we first generate multi-view images simultaneously using a grid representation to enhance global understanding. Meanwhile, we propose a view synchronization and adaptive weighting module during diffusion sampling to further ensure local consistency. Finally, a 3D-aware texture completion model combined with a texture enhancement model is used to generate seamless, high-resolution texture maps. Comprehensive experiments demonstrate that our framework significantly outperforms state-of-the-art methods in both flexibility and generation quality. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:36:03 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Yan",
"Dongyu",
""
],
[
"Wu",
"Leyi",
""
],
[
"Lin",
"Jiantao",
""
],
[
"Wang",
"Luozhou",
""
],
[
"Xu",
"Tianshuo",
""
],
[
"Chen",
"Zhifei",
""
],
[
"Yang",
"Zhen",
""
],
[
"Xu",
"Lie",
""
],
[
"Zhang",
"Shunsi",
""
],
[
"Chen",
"Yingcong",
""
]
] |
2506.02621 | Zhaoyang Li | Zhaoyang Li, Haodong Zhou, Longjie Luo, Xiaoxiao Li, Yongxin Chen, Lin Li, Qingyang Hong | Cross-attention and Self-attention for Audio-visual Speaker Diarization in MISP-Meeting Challenge | null | null | null | null | cs.SD | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper presents the system developed for Task 1 of the Multi-modal Information-based Speech Processing (MISP) 2025 Challenge. We introduce CASA-Net, an embedding fusion method designed for end-to-end audio-visual speaker diarization (AVSD) systems. CASA-Net incorporates a cross-attention (CA) module to effectively capture cross-modal interactions in audio-visual signals and employs a self-attention (SA) module to learn contextual relationships among audio-visual frames. To further enhance performance, we adopt a training strategy that integrates pseudo-label refinement and retraining, improving the accuracy of timestamp predictions. Additionally, median filtering and overlap averaging are applied as post-processing techniques to eliminate outliers and smooth prediction labels. Our system achieved a diarization error rate (DER) of 8.18% on the evaluation set, representing a relative improvement of 47.3% over the baseline DER of 15.52%. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:38:05 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Li",
"Zhaoyang",
""
],
[
"Zhou",
"Haodong",
""
],
[
"Luo",
"Longjie",
""
],
[
"Li",
"Xiaoxiao",
""
],
[
"Chen",
"Yongxin",
""
],
[
"Li",
"Lin",
""
],
[
"Hong",
"Qingyang",
""
]
] |
2506.02622 | Carmine Tommaso Recchiuto | Omotoye Shamsudeen Adekoya, Antonio Sgorbissa, Carmine Tommaso Recchiuto | HORUS: A Mixed Reality Interface for Managing Teams of Mobile Robots | 7 pages, 7 figures, conference paper submitted to IROS 2025 | null | null | null | cs.RO cs.HC | http://creativecommons.org/licenses/by/4.0/ | Mixed Reality (MR) interfaces have been extensively explored for controlling mobile robots, but there is limited research on their application to managing teams of robots. This paper presents HORUS: Holistic Operational Reality for Unified Systems, a Mixed Reality interface offering a comprehensive set of tools for managing multiple mobile robots simultaneously. HORUS enables operators to monitor individual robot statuses, visualize sensor data projected in real time, and assign tasks to single robots, subsets of the team, or the entire group, all from a Mini-Map (Ground Station). The interface also provides different teleoperation modes: a mini-map mode that allows teleoperation while observing the robot model and its transform on the mini-map, and a semi-immersive mode that offers a flat, screen-like view in either single or stereo view (3D). We conducted a user study in which participants used HORUS to manage a team of mobile robots tasked with finding clues in an environment, simulating search and rescue tasks. This study compared HORUS's full-team management capabilities with individual robot teleoperation. The experiments validated the versatility and effectiveness of HORUS in multi-robot coordination, demonstrating its potential to advance human-robot collaboration in dynamic, team-based environments. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:38:46 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Adekoya",
"Omotoye Shamsudeen",
""
],
[
"Sgorbissa",
"Antonio",
""
],
[
"Recchiuto",
"Carmine Tommaso",
""
]
] |
2506.02623 | Yuyang Zhou | Yuyang Zhou, Ferrante Neri, Yew-Soon Ong, Ruibin Bai | SiamNAS: Siamese Surrogate Model for Dominance Relation Prediction in Multi-objective Neural Architecture Search | Genetic and Evolutionary Computation Conference (GECCO' 25) | null | null | null | cs.LG cs.AI cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Modern neural architecture search (NAS) is inherently multi-objective, balancing trade-offs such as accuracy, parameter count, and computational cost. This complexity makes NAS computationally expensive and nearly impossible to solve without efficient approximations. To address this, we propose a novel surrogate modelling approach that leverages an ensemble of Siamese network blocks to predict dominance relationships between candidate architectures. Lightweight and easy to train, the surrogate achieves 92% accuracy and replaces the crowding distance calculation in the survivor selection strategy with a heuristic rule based on model size. Integrated into a framework termed SiamNAS, this design eliminates costly evaluations during the search process. Experiments on NAS-Bench-201 demonstrate the framework's ability to identify Pareto-optimal solutions with significantly reduced computational costs. The proposed SiamNAS identified a final non-dominated set containing the best architecture in NAS-Bench-201 for CIFAR-10 and the second-best for ImageNet, in terms of test error rate, within 0.01 GPU days. This proof-of-concept study highlights the potential of the proposed Siamese network surrogate model to generalise to multi-tasking optimisation, enabling simultaneous optimisation across tasks. Additionally, it offers opportunities to extend the approach for generating Sets of Pareto Sets (SOS), providing diverse Pareto-optimal solutions for heterogeneous task settings. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:39:42 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Zhou",
"Yuyang",
""
],
[
"Neri",
"Ferrante",
""
],
[
"Ong",
"Yew-Soon",
""
],
[
"Bai",
"Ruibin",
""
]
] |
2506.02624 | Taufik Abrao PhD | Bruno Felipe Costa and Taufik Abr\~ao | Refined Metrics, Sensing Limits, and Resource Allocation in OTFS-RSMA LEO ISAC | 6 pages, 2 figures, 5 graphs, conference paper | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | This paper develops an integrated OTFS-RSMA framework employing advanced SP techniques tailored for this demanding environment. We derive refined communication performance metrics, specifically SINR expressions capturing the practical effects of ICSI and ISIC. Moreover, fundamental sensing limits are established via CRB derivation incorporating parameter-dependent echo gain, linking waveform SP properties to estimation accuracy. The resource allocation is formulated as a non-convex optimization problem aiming for Max-Min Fairness under constraints derived from these SP metrics. Illustrative results, obtained via GA optimization, crucially demonstrate that the proposed RSMA scheme uniquely enables the simultaneous satisfaction of stringent communication and sensing constraints metrics, a capability not achieved by conventional SDMA. Such results {highlight the efficacy of the integrated OTFS-RSMA precoding and optimization approach for designing robust and feasible LEO-ISAC systems.
Index Terms -- ISAC, LEO, OTFS, RSMA, Channel Modeling, CRB, SINR, ICSI, ISIC, Resource Allocation, Max-Min Fairness, Delay-Doppler (DD) Processing, Satellite Communications. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:40:14 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Costa",
"Bruno Felipe",
""
],
[
"Abrão",
"Taufik",
""
]
] |
2506.02625 | Hongjian Sun Prof. | Ahmad Massud Tota Khel, Aissa Ikhlef, Zhiguo Ding, and Hongjian Sun | Zero-Energy RIS-Assisted Communications With Noise Modulation and Interference-Based Energy Harvesting | 32 pages, 12 figures, accepted by IEEE Transactions on Green Communications and Networking | null | null | null | cs.IT cs.ET cs.NI math.IT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | To advance towards carbon-neutrality and improve the limited {performance} of conventional passive wireless communications, in this paper, we investigate the integration of noise modulation with zero-energy reconfigurable intelligent surfaces (RISs). In particular, the RIS reconfigurable elements (REs) are divided into two groups: one for beamforming the desired signals in reflection mode and another for harvesting energy from interference signals in an absorption mode, providing the power required for RIS operation. Since the harvested energy is a random variable, a random number of REs can beamform the signals, while the remainder blindly reflects them. We present a closed-form solution and a search algorithm for REs allocation, jointly optimizing both the energy harvesting (EH) and communication performance. Considering the repetition coding technique and discrete phase shifts, we derive analytical expressions for the energy constrained success rate, bit error rate, optimal threshold, mutual information, {and energy efficiency}. Numerical and simulation results confirm the effectiveness of the algorithm and expressions, demonstrating the superiority of the proposed integration over conventional noise-modulation systems. It is shown that by properly allocating the REs, both the EH and communication performance can be improved in low to moderate interference scenarios, while the latter is restricted in the high-interference regime. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:40:24 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Khel",
"Ahmad Massud Tota",
""
],
[
"Ikhlef",
"Aissa",
""
],
[
"Ding",
"Zhiguo",
""
],
[
"Sun",
"Hongjian",
""
]
] |
2506.02626 | Ada Sawilska | Ada Sawilska, Mateusz Trokielewicz | Synthetic Iris Image Databases and Identity Leakage: Risks and Mitigation Strategies | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a comprehensive overview of iris image synthesis methods, which can alleviate the issues associated with gathering large, diverse datasets of biometric data from living individuals, which are considered pivotal for biometric methods development. These methods for synthesizing iris data range from traditional, hand crafted image processing-based techniques, through various iterations of GAN-based image generators, variational autoencoders (VAEs), as well as diffusion models. The potential and fidelity in iris image generation of each method is discussed and examples of inferred predictions are provided. Furthermore, the risks of individual biometric features leakage from the training sets are considered, together with possible strategies for preventing them, which have to be implemented should these generative methods be considered a valid replacement of real-world biometric datasets. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:41:43 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Sawilska",
"Ada",
""
],
[
"Trokielewicz",
"Mateusz",
""
]
] |
2506.02627 | \"Omer Tarik \"Ozyilmaz | \"Omer Tarik \"Ozyilmaz, Matt Coler, Matias Valdenegro-Toro | Overcoming Data Scarcity in Multi-Dialectal Arabic ASR via Whisper Fine-Tuning | Accepted at Interspeech 2025 | null | null | null | cs.CL cs.SD eess.AS | http://creativecommons.org/publicdomain/zero/1.0/ | Although commercial Arabic automatic speech recognition (ASR) systems support Modern Standard Arabic (MSA), they struggle with dialectal speech. We investigate the effect of fine-tuning OpenAI's Whisper on five major Arabic dialects (Gulf, Levantine, Iraqi, Egyptian, Maghrebi) using Mozilla Common Voice for MSA and the MASC dataset for dialectal speech. We evaluate MSA training size effects, benefits of pre-training on MSA data, and dialect-specific versus dialect-pooled models. We find that small amounts of MSA fine-tuning data yield substantial improvements for smaller models, matching larger non-fine-tuned models. While MSA pre-training shows minimal benefit, suggesting limited shared features between MSA and dialects, our dialect-pooled models perform comparably to dialect-specific ones. This indicates that pooling dialectal data, when properly balanced, can help address data scarcity in low-resource ASR without significant performance loss. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:41:49 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Özyilmaz",
"Ömer Tarik",
""
],
[
"Coler",
"Matt",
""
],
[
"Valdenegro-Toro",
"Matias",
""
]
] |
2506.02628 | Frank Stefani | F. Stefani, T. Weier, G.M. Horstmann, G. Mamatsashvili | Reappraising the Elatina series: Solar dynamo clocking and inference of orbital periods | 10 pages, 7 figures | null | null | null | astro-ph.SR astro-ph.EP | http://creativecommons.org/licenses/by/4.0/ | We reconsider the 680 million year old Elatina series of sedimentary laminae from South Australia that show a remarkably stable periodicity with a main period of around 12 years, which is close to the Schwabe cycle, and a second period of 314 years that has been coined Elatina cycle. By analyzing the residuals of the series' minima from a linear trend, and deriving Dicke's ratio, we first show that the series exhibits a high degree of phase stability, except one single break point which may indicate a 90{\deg} phase jump. We discuss the data in terms of a recently developed synchronization model of the solar dynamo. This model is then employed to infer those orbital periods of Venus, Earth, Jupiter and Saturn that would be required to jointly explain the moderately changed Schwabe cycle, and the Elatina cycle when interpreted as a prolonged Suess-de Vries cycle. Assuming pairwise conservations of the sum of the angular momenta of Jupiter/Saturn and Venus/Earth, respectively, we find solutions of the underlying inverse problem which amount to approximately 1 percent angular momentum increase of Jupiter and a 0.005 per cent angular momentum increase of Earth. The plausibility of such changes over a period of seven hundred million years is discussed in light of solar system dynamics. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:44:09 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Stefani",
"F.",
""
],
[
"Weier",
"T.",
""
],
[
"Horstmann",
"G. M.",
""
],
[
"Mamatsashvili",
"G.",
""
]
] |
2506.02629 | Hakk{\i} Tuncay Ozer | H.T. \"Ozer, Ayt\"ul Filiz | $\mathcal{N}=1$ Jackiw -Teitelboim supergravity beyond the Schwarzian regime | 29 pages,LaTeX | null | null | null | hep-th gr-qc math-ph math.MP | http://creativecommons.org/publicdomain/zero/1.0/ | We investigate the asymptotic symmetry structure of two--dimensional dilaton gravity in its $\mathcal{N}=1$ supersymmetric extension based on the $\mathfrak{osp}(1|2)$ Lie superalgebra. Within the BF theoretical framework, we analyze affine and superconformal boundary conditions for each case and systematically derive the associated asymptotic symmetry algebras. While the classical theory recovers the Virasoro algebra or its affine enhancement, the supersymmetric extension yields a classical $\mathcal{N}=1$ superconformal algebra, subject to dynamical symmetry breaking mechanisms induced by the dilaton supermultiplet. We find that the boundary behavior of the dilaton not only leads to a reduction of the full affine $\mathfrak{osp}(1|2)_k$ symmetry down to $\tt{O} \tt{S} p(1|2)$, but also introduces an abelian extension through commuting modes. These results reveal a novel interplay between symmetry breaking and symmetry extension in low-dimensional supergravity. Our construction generalizes previous analyses of $sl(2,\mathbb{R})$ dilaton gravity to the supersymmetric domain and offers a consistent foundation for investigating boundary dynamics beyond the Schwarzian regime. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:45:46 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Özer",
"H. T.",
""
],
[
"Filiz",
"Aytül",
""
]
] |
2506.02630 | Tom Jacobs | Tom Jacobs, Advait Gadhikar, Celia Rubio-Madrigal, and Rebekka Burkholz | HAM: A Hyperbolic Step to Regulate Implicit Bias | 26 pages, 7 figures | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Understanding the implicit bias of optimization algorithms has become central to explaining the generalization behavior of deep learning models. For instance, the hyperbolic implicit bias induced by the overparameterization $m \odot w$--though effective in promoting sparsity--can result in a small effective learning rate, which slows down convergence. To overcome this obstacle, we propose HAM (Hyperbolic Aware Minimization), which alternates between an optimizer step and a new hyperbolic mirror step. We derive the Riemannian gradient flow for its combination with gradient descent, leading to improved convergence and a similar beneficial hyperbolic geometry as $m \odot w$ for feature learning. We provide an interpretation of the the algorithm by relating it to natural gradient descent, and an exact characterization of its implicit bias for underdetermined linear regression. HAM's implicit bias consistently boosts performance--even of dense training, as we demonstrate in experiments across diverse tasks, including vision, graph and node classification, and large language model fine-tuning. HAM is especially effective in combination with different sparsification methods, improving upon the state of the art. The hyperbolic step requires minimal computational and memory overhead, it succeeds even with small batch sizes, and its implementation integrates smoothly with existing optimizers. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:47:16 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Jacobs",
"Tom",
""
],
[
"Gadhikar",
"Advait",
""
],
[
"Rubio-Madrigal",
"Celia",
""
],
[
"Burkholz",
"Rebekka",
""
]
] |
2506.02631 | Antoine Lotz | Thomas Deschatre, Pierre Gruet, Antoine Lotz | A non-local estimator for locally stationary Hawkes processes | null | null | null | null | math.ST stat.TH | http://creativecommons.org/licenses/by/4.0/ | We consider the problem of estimating the parameters of a non-stationary Hawkes process with time-dependent reproduction rate and baseline intensity. Our approach relies on the standard maximum likelihood estimator (MLE), coinciding with the conventional approach for stationary point processes characterised by [Ogata, 1978]. In the fully parametric setting, we find that the MLE over a single observation of the process over $[0, T]$ remains consistent and asymptotically normal as $T \to \infty$. Our results extend partially to the semi-nonparametric setting where no specific shape is assumed for the reproduction rate $g \colon [0, 1] \mapsto \mathbb{R}_+$. We construct a time invariance test with null hypothesis that g is constant against the alternative that it is not, and find that it remains consistent over the whole space of continuous functions of [0, 1]. As an application, we employ our procedure in the context of the German intraday power market, where we provide evidence of fluctuations in the endogeneity rate of the order flow. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:47:26 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Deschatre",
"Thomas",
""
],
[
"Gruet",
"Pierre",
""
],
[
"Lotz",
"Antoine",
""
]
] |
2506.02632 | Katsiaryna Pantsialeyeva | Elena Melnikova, Katsiaryna Pantsialeyeva, Dmitry Gorbach, Alexei Tolstik, Sergei Slussarenko Jr., Alina Karabchevsky | Phase Topology Stability of an Optical Vortex via an Electrically Controlled Twist-Planar Oriented Liquid Crystal Fresnel Lens | null | null | null | null | physics.optics | http://creativecommons.org/licenses/by/4.0/ | Optical vortices (OVs) have emerged as a revolutionary concept in modern photonics, offering a unique method of manipulating light beyond conventional Gaussian beams. Despite their vast potential, phase topology stability remains unaddressed, limiting their widespread adoption and performance in real-world environments. Here, we reveal the missing link to assessing the stability of optical vortices using an electrically tunable twist-planar liquid crystal (LC) Fresnel lens. The proposed LC-based lens leverages the birefringence and voltage-controlled reconfigurability of liquid crystals to dynamically probe the phase topology of singular beams. By modulating the LC orientation with an applied voltage, we restructure the optical phase in real-time without requiring modifications to the optical setup. The 3V and 35V voltage supply allows for the switch between the "topological charge detection" and "optical singular beam propagation" modes. This eliminates the need for additional optical elements, significantly simplifying the detection and characterization of vortex beams. Experimental and theoretical investigations demonstrate that the vortex topology can be unambiguously identified from the intensity profile observed in the Fourier plane of a lens. Furthermore, the designed device features low power consumption, compact form factor, and seamless integration potential, making it a promising candidate for scalable vortex-based photonic systems. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:48:26 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Melnikova",
"Elena",
""
],
[
"Pantsialeyeva",
"Katsiaryna",
""
],
[
"Gorbach",
"Dmitry",
""
],
[
"Tolstik",
"Alexei",
""
],
[
"Slussarenko",
"Sergei",
"Jr."
],
[
"Karabchevsky",
"Alina",
""
]
] |
2506.02633 | Cheng Yang | Cheng Yang and Lijing Liang and Zhixun Su | ControlMambaIR: Conditional Controls with State-Space Model for Image Restoration | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper proposes ControlMambaIR, a novel image restoration method designed to address perceptual challenges in image deraining, deblurring, and denoising tasks. By integrating the Mamba network architecture with the diffusion model, the condition network achieves refined conditional control, thereby enhancing the control and optimization of the image generation process. To evaluate the robustness and generalization capability of our method across various image degradation conditions, extensive experiments were conducted on several benchmark datasets, including Rain100H, Rain100L, GoPro, and SSID. The results demonstrate that our proposed approach consistently surpasses existing methods in perceptual quality metrics, such as LPIPS and FID, while maintaining comparable performance in image distortion metrics, including PSNR and SSIM, highlighting its effectiveness and adaptability. Notably, ablation experiments reveal that directly noise prediction in the diffusion process achieves better performance, effectively balancing noise suppression and detail preservation. Furthermore, the findings indicate that the Mamba architecture is particularly well-suited as a conditional control network for diffusion models, outperforming both CNN- and Attention-based approaches in this context. Overall, these results highlight the flexibility and effectiveness of ControlMambaIR in addressing a range of image restoration perceptual challenges. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:50:00 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Yang",
"Cheng",
""
],
[
"Liang",
"Lijing",
""
],
[
"Su",
"Zhixun",
""
]
] |
2506.02634 | Jiahao Wang | Jiahao Wang, Jinbo Han, Xingda Wei, Sijie Shen, Dingyan Zhang, Chenguang Fang, Rong Chen, Wenyuan Yu, Haibo Chen | KVCache Cache in the Wild: Characterizing and Optimizing KVCache Cache at a Large Cloud Provider | Accepted by USENIX ATC'25 | null | null | null | cs.DC cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Serving large language models (LLMs) is important for cloud providers, and caching intermediate results (KV\$) after processing each request substantially improves serving throughput and latency. However, there is limited understanding of how LLM serving benefits from KV\$ caching, where system design decisions like cache eviction policies are highly workload-dependent. In this paper, we present the first systematic characterization of the KV\$ workload patterns from one of the leading LLM service providers. We draw observations that were not covered by previous studies focusing on synthetic workloads, including: KV\$ reuses are skewed across requests, where reuses between single-turn requests are equally important as multi-turn requests; the reuse time and probability are diverse considering all requests, but for a specific request category, the pattern tends to be predictable; and the overall cache size required for an ideal cache hit ratio is moderate. Based on the characterization, we further propose a workload-aware cache eviction policy that improves the serving performance under real-world traces, especially with limited cache capacity. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:51:38 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Wang",
"Jiahao",
""
],
[
"Han",
"Jinbo",
""
],
[
"Wei",
"Xingda",
""
],
[
"Shen",
"Sijie",
""
],
[
"Zhang",
"Dingyan",
""
],
[
"Fang",
"Chenguang",
""
],
[
"Chen",
"Rong",
""
],
[
"Yu",
"Wenyuan",
""
],
[
"Chen",
"Haibo",
""
]
] |
2506.02636 | Matthias Zach | Matthias Zach | Some computational aspects of spectral sequences in \v{C}ech cohomology | 17 pages, 1 figure | null | null | null | math.AG math.AC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Sheaf cohomology or, more generally, higher direct images of coherent sheaves along proper morphisms are central to modern algebraic geometry. However, the computation of these objects is a non-trivial and expensive task which easily challenges the capacities of modern computers. We describe an algorithm and its implementation to compute a spectral sequence converging to the higher direct images of a bounded complex of sheaves on a product of projective spaces $\mathbb P = \mathbb P^{r_1}\times \dots \times \mathbb P^{r_m}$ over an arbitrary affine base $\mathrm{Spec} R$. We assume the ring $R$ to be computable and the complex of sheaves to be represented by an actual complex of (multi-)graded modules. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:53:29 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Zach",
"Matthias",
""
]
] |
2506.02637 | Louis Vervoort | Konstantinos Papatryfonos, Louis Vervoort | Proposed experiments for detecting contextual hidden variables | null | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose two quantum experiments - modified Bell tests - that could detect contextual hidden variables underlying quantum mechanics. The experiments are inspired by hydrodynamic pilot-wave systems that mimic a wide range of quantum effects and exhibit a classical analog of contextuality. To justify the experiments, we show that contextual hidden variables are 'physics as usual' if a unification between quantum mechanics and general relativity is possible. Accordingly, contextual theories can bypass Bell's theorem in a way that is both local and non-conspiratorial. We end with a note on the relevance of exploratory experiments in the foundations of quantum physics. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:53:39 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Papatryfonos",
"Konstantinos",
""
],
[
"Vervoort",
"Louis",
""
]
] |
2506.02638 | Shang Li | Shang Li | Toroidal embedding of Chevalley groups over $\mathbb{Z}$ | Comments are welcome | null | null | null | math.AG | http://creativecommons.org/licenses/by/4.0/ | The classification of equivariant toroidal embeddings of a reductive group over an algebraically closed field is combinatorial and does not depend on the characteristic of the base field. This suggests that there should exist ``universal'' toroidal embeddings for a Chevalley group scheme over $\mathbb{Z}$ which specialize to classical toroidal embeddings via base change. In this paper, we establish the existence of ``universal'' equivariant toroidal embeddings for split reductive group schemes over $\mathbb{Z}$. We also discuss several geometric properties of these embeddings. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:53:56 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Li",
"Shang",
""
]
] |
2506.02639 | Hiroshi Okada | Takaaki Nomura, Hiroshi Okada | Neutrino mass model at a three-loop level from a non-holomorphic modular $A_4$ symmetry | 17 pages, 5 tables, 8 figures. arXiv admin note: text overlap with arXiv:2412.18095 | null | null | null | hep-ph | http://creativecommons.org/publicdomain/zero/1.0/ | We study a three-loop induced neutrino mass scenario from a non-holomorphic modular $A_4$ flavor symmetry and reach the minimum scenario leading predictions of the lepton masses, mixing angles, and Dirac and Majorana phases, which are shown through the chi square analyses. In addition, we discuss the lepton flavor violations, muon anomalous magnetic moment, lepton universality, and relic density of the dark matter candidate. And, we find that we need to extend our model if we satisfy the observed relic density of dark matter within the limit of perturbation where it can be done by adding one singlet scalar boson without changing predictions in neutrino sector. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:53:56 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Nomura",
"Takaaki",
""
],
[
"Okada",
"Hiroshi",
""
]
] |
2506.02640 | Jonas Lippold | Uta Freiberg, Jonas Lippold | A family of non Minkowski measurable fractals in $\mathbb{R}^2$ | null | null | null | null | math.MG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A long-standing conjecture of Lapidus asserts that, under certain conditions, a self-similar fractal set is not Minkowski measurable if and only if it is of lattice-type. For self-similar sets in $\mathbb{R}$, the Lapidus conjecture has been confirmed. However, in higher dimensions, it remains unclear whether all lattice-type self-similar sets are not Minkowski measurable. This work presents a family of lattice-type subsets in $\mathbb{R}^2$ that are not Minkowski measurable, hence providing further support for the conjecture. Furthermore, an argument is presented to illustrate why these sets are not covered by previous results. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:54:27 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Freiberg",
"Uta",
""
],
[
"Lippold",
"Jonas",
""
]
] |
2506.02641 | Charlotte Vastel | Charlotte Vastel and Francesco Fontani | Observations of non complex organic molecules in the gas phase of the interstellar medium | Accepted in Elsevier Handbook of Astrochemistry (2025) | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | The field of astrochemistry has seen major advances triggered by the completion of new powerful radio telescopes, with gains in sensitivity of receivers and in bandwidth. To date, about 330 molecular species are detected, in interstellar clouds, circumstellar shells and even extragalactic sources. The first interstellar molecules were first discovered through their electronic transitions in the visual and near UV regions of the spectra in the 1930s. Then the discovery of (pure) rotational transitions of interstellar molecules dates back to the late 1960s. The improvement of detectors and the increase in telescope sizes really opened up the submillimeter sky. The radio and submillimeter ranges cover the lowest rotational lines of molecular species. The bigger the molecule, the more spectral lines at different frequencies it produces, with weaker line intensities. Over the past 30 years, we have discovered that we live in a molecular universe, where molecules are abundant and widespread, probing the structure and evolution of galaxies, as well as the temperature and density of the observed medium, opening a new field called astrochemistry. The progress has been dramatic, since the discovery of the first molecules about 100 years ago. We present in this review, the detection techniques that led to the discovery of the simple molecules in the gas phase and the methodology that lead to the abundances determinations and the comparison with chemical modelling. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:56:59 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Vastel",
"Charlotte",
""
],
[
"Fontani",
"Francesco",
""
]
] |
2506.02642 | Hongjian Sun Prof. | Huijun Tang, Jieling Zhang, Zhidong Zhao, Huaming Wu, Hongjian Sun, and Pengfei Jiao | Joint Optimization based on Two-phase GNN in RIS- and DF-assisted MISO Systems with Fine-grained Rate Demands | 14 Pages, 9 figures, accepted by IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS | null | null | null | cs.IT eess.SP math.IT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Reconfigurable intelligent Surfaces (RIS) and half-duplex decoded and forwarded (DF) relays can collaborate to optimize wireless signal propagation in communication systems. Users typically have different rate demands and are clustered into groups in practice based on their requirements, where the former results in the trade-off between maximizing the rate and satisfying fine-grained rate demands, while the latter causes a trade-off between inter-group competition and intra-group cooperation when maximizing the sum rate. However, traditional approaches often overlook the joint optimization encompassing both of these trade-offs, disregarding potential optimal solutions and leaving some users even consistently at low date rates. To address this issue, we propose a novel joint optimization model for a RIS- and DF-assisted multiple-input single-output (MISO) system where a base station (BS) is with multiple antennas transmits data by multiple RISs and DF relays to serve grouped users with fine-grained rate demands. We design a new loss function to not only optimize the sum rate of all groups but also adjust the satisfaction ratio of fine-grained rate demands by modifying the penalty parameter. We further propose a two-phase graph neural network (GNN) based approach that inputs channel state information (CSI) to simultaneously and autonomously learn efficient phase shifts, beamforming, and relay selection. The experimental results demonstrate that the proposed method significantly improves system performance. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:57:25 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Tang",
"Huijun",
""
],
[
"Zhang",
"Jieling",
""
],
[
"Zhao",
"Zhidong",
""
],
[
"Wu",
"Huaming",
""
],
[
"Sun",
"Hongjian",
""
],
[
"Jiao",
"Pengfei",
""
]
] |
2506.02643 | Ju-Feng Wu | Hansheng Diao, Giovanni Rosso, Ju-Feng Wu | Overconvergent Eichler-Shimura morphisms for $\mathrm{GSp}_4$ | null | null | null | null | math.NT math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We construct explicit Eichler-Shimura morphisms for families of overconvergent Siegel modular forms of genus two. These can be viewed as $p$-adic interpolations of the Eichler-Shimura decomposition of Faltings-Chai for classical Siegel modular forms. In particular, we are able to $p$-adically interpolate the entire decomposition, extending our previous work on the $H^0$-part. The key new inputs are the higher Coleman theory of Boxer-Pilloni and a theory of pro-Kummer \'etale cohomology with supports. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:57:56 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Diao",
"Hansheng",
""
],
[
"Rosso",
"Giovanni",
""
],
[
"Wu",
"Ju-Feng",
""
]
] |
2506.02645 | Oded Farago | R. K. Singh, Oded Farago | Steady state and relaxation dynamics of run and tumble particles in contact with a heat bath | 22 pages, 6 figures. Accepted for publication in Phys. Rev. E | null | null | null | cond-mat.stat-mech cond-mat.soft | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the relaxation dynamics of a run and tumble particle in a one-dimensional piecewise linear potential $U(x)=b|x|$, from delta-function initial conditions at $x=0$ to steady state. In addition to experiencing active telegraphic noise, the particle is in contact with a heat bath at temperature $T$ that applies white thermal noise. We find that the position distribution of the RTP is described by a sum of two distributions ("modes"), each of which of the form $P(x,t\to\infty)\sim e^{-\lambda_i|x|}$ ($i=1,2$) at steady state. The two modes are dynamically coupled: At very short times ($t\to 0$), each mode stores half of the probability, and exhibits thermal diffusive spreading with a Gaussian profile. With progressing time and evolution toward steady state, the partition of probability between the modes becomes increasingly uneven and, depending on the model parameters, the mode with the smaller value of $\lambda_i$ may carry an overwhelming majority of the probability. Moreover, we identify that the characteristic relaxation time of each mode is $\tau_i=(\lambda_i^2T)^{-1}$, which implies that the minority mode also relaxes much faster than the dominant one. A more detailed analysis reveals that $\tau_i$ is characteristic of the mode relaxation only close to the origin at the core of the distribution, while further away it increases linearly with $|x|$ as if a relaxation front is propagating at constant speed $v_i^*=2\sqrt{T/\tau_i}$ in the system. The rate of non-equilibrium entropy production can be related to the two-mode splitting of the probability distribution and be expressed in terms of their correlation-lengths $\lambda_i$ and their contributions to the steady state distribution. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:00:06 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Singh",
"R. K.",
""
],
[
"Farago",
"Oded",
""
]
] |
2506.02646 | Sabah Al-Fedaghi Dr. | Sabah Al-Fedaghi | Textual-Based vs. Thinging Machines Conceptual Modeling | 12 pages, 20 figures | null | null | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | Software engineers typically interpret the domain description in natural language and translate it into a conceptual model. Three approaches are used in this domain modeling: textual languages, diagrammatic languages, and a mixed based of text and diagrams. According to some researchers, relying on a diagrammatic notation levies certain burdens for designing large models because visual languages are intended to depict everything diagrammatically during a development process but fail to do so for a lack of developer efficiency. It is claimed that textual formats enable easier manipulation in editors and tools and facilitate the integration of ontologies in software systems. In this paper, we explore the problem of the relationship between textual format and diagramming in conceptual modeling. The main focus is modeling based on the so-called thinging machine (TM). Several examples are developed in detail to contrast side-by-side targeted domains represented in textual description and TM modeling. A TM model is defined as a thimac (thing/machine) with a time feature that forms dynamic events over static thimacs utilizing five generic actions: create, process, release, transfer, and receive. This provides a conceptual foundation that can be simplified further by eliminating the actions of release, transfer, and receive. A multilevel reduction in the TM diagram s complexity can also be achieved by assuming diagrammatic notations represent the actions of creation and processing. We envision that special tools will help improve developer efficiency. The study s results of contrasting textual and mix-based descriptions vs. TM modeling justify our claim that TM modeling is a more appropriate methodology than other diagrammatic schemes (e.g., UML classes) examined in this paper. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:00:26 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Al-Fedaghi",
"Sabah",
""
]
] |
2506.02647 | Niklas Baumgarten | Niklas Baumgarten, David Schneiderhan | Multilevel Stochastic Gradient Descent for Optimal Control Under Uncertainty | null | null | null | null | math.OC cs.MS cs.NA math.NA | http://creativecommons.org/licenses/by-sa/4.0/ | We present a multilevel stochastic gradient descent method for the optimal control of systems governed by partial differential equations under uncertain input data. The gradient descent method used to find the optimal control leverages a parallel multilevel Monte Carlo method as stochastic gradient estimator. As a result, we achieve precise control over the stochastic gradient's bias, introduced by numerical approximation, and its sampling error, arising from the use of incomplete gradients, while optimally managing computational resources. We show that the method exhibits linear convergence in the number of optimization steps while avoiding the cost of computing the full gradient at the highest fidelity. Numerical experiments demonstrate that the method significantly outperforms the standard (mini-) batched stochastic gradient descent method in terms of convergence speed and accuracy. The method is particularly well-suited for high-dimensional control problems, taking advantage of parallel computing resources and a distributed multilevel data structure. Additionally, we evaluate and implement different step size strategies, optimizer schemes, and budgeting techniques. The method's performance is studied using a two-dimensional elliptic subsurface diffusion problem with log-normal coefficients and Mat\'ern covariance. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:00:43 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Baumgarten",
"Niklas",
""
],
[
"Schneiderhan",
"David",
""
]
] |
2506.02648 | Yue Yang | Yue Yang, MingKang Chen, Qihua Liu, Mengkang Hu, Qiguang Chen, Gengrui Zhang, Shuyue Hu, Guangtao Zhai, Yu Qiao, Yu Wang, Wenqi Shao, Ping Luo | Truly Assessing Fluid Intelligence of Large Language Models through Dynamic Reasoning Evaluation | null | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advances in large language models (LLMs) have demonstrated impressive reasoning capacities that mirror human-like thinking. However, whether LLMs possess genuine fluid intelligence (i.e., the ability to reason abstractly and generalize rules in novel situations) remains an open question. Existing reasoning benchmarks either focus on domain-specific knowledge (crystallized intelligence) or lack interpretability. To address these limitations, we propose DRE-Bench, a dynamic reasoning evaluation benchmark grounded in a hierarchical cognitive framework. DRE-Bench consists of 36 abstract reasoning tasks organized across four cognitive levels, with each task featuring multiple dynamic variants that test the same underlying latent rule. This design enables fine-grained, interpretable, and reliable assessments of fluid intelligence. We evaluate a range of state-of-the-art LLMs, including both general LLMs (GPT-4o, Claude 3.7) and reasoning LLMs (o1, DeepSeek-R1, QwQ, Skywork-OR1). Experimental results reveal that although most LLMs achieve competent and robust performance in low-level cognition, they struggle with high-level cognition and exhibit limited generalization as task complexity grows. Our findings highlight the gap between current LLMs and true human-like fluid intelligence and offer a new path for systematically tracking reasoning progress in LLMs. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:01:08 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Yang",
"Yue",
""
],
[
"Chen",
"MingKang",
""
],
[
"Liu",
"Qihua",
""
],
[
"Hu",
"Mengkang",
""
],
[
"Chen",
"Qiguang",
""
],
[
"Zhang",
"Gengrui",
""
],
[
"Hu",
"Shuyue",
""
],
[
"Zhai",
"Guangtao",
""
],
[
"Qiao",
"Yu",
""
],
[
"Wang",
"Yu",
""
],
[
"Shao",
"Wenqi",
""
],
[
"Luo",
"Ping",
""
]
] |
2506.02649 | Yousef Emami | Yousef Emami, Hao Zhou, Miguel Gutierrez Gaitan, Kai Li, Luis Almeida, Zhu Han | From Prompts to Protection: Large Language Model-Enabled In-Context Learning for Smart Public Safety UAV | 8 pages, 4 figures | null | null | null | cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | A public safety Unmanned Aerial Vehicle (UAV) enhances situational awareness in emergency response. Its agility and ability to optimize mobility and establish Line-of-Sight (LoS) communication make it increasingly vital for managing emergencies such as disaster response, search and rescue, and wildfire monitoring. While Deep Reinforcement Learning (DRL) has been applied to optimize UAV navigation and control, its high training complexity, low sample efficiency, and simulation-to-reality gap limit its practicality in public safety. Recent advances in Large Language Models (LLMs) offer a compelling alternative. With strong reasoning and generalization capabilities, LLMs can adapt to new tasks through In-Context Learning (ICL), which enables task adaptation via natural language prompts and example-based guidance, without retraining. Deploying LLMs at the network edge, rather than in the cloud, further reduces latency and preserves data privacy, thereby making them suitable for real-time, mission-critical public safety UAVs. This paper proposes the integration of LLM-enabled ICL with public safety UAV to address the key functions, such as path planning and velocity control, in the context of emergency response. We present a case study on data collection scheduling where the LLM-enabled ICL framework can significantly reduce packet loss compared to conventional approaches, while also mitigating potential jailbreaking vulnerabilities. Finally, we discuss LLM optimizers and specify future research directions. The ICL framework enables adaptive, context-aware decision-making for public safety UAV, thus offering a lightweight and efficient solution for enhancing UAV autonomy and responsiveness in emergencies. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:01:33 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Emami",
"Yousef",
""
],
[
"Zhou",
"Hao",
""
],
[
"Gaitan",
"Miguel Gutierrez",
""
],
[
"Li",
"Kai",
""
],
[
"Almeida",
"Luis",
""
],
[
"Han",
"Zhu",
""
]
] |
2506.02650 | Shukun Wu | Shukun Wu | Weighted $L^2$ estimates with applications to $L^p$ problems | null | null | null | null | math.CA math.AP | http://creativecommons.org/licenses/by/4.0/ | We establish some weighted $L^2$ estimates for the Fourier extension operator in $\mathbb{R}^2$ and discuss several applications to $L^p$ problems. These include estimates for the maximal Schr\"odinger operator and the maximal extension operator, decay of circular $L^p$-means of Fourier transform of fractal measures, and an $L^p$ analogue of the Mizohata-Takeuchi conjecture. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:02:12 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Wu",
"Shukun",
""
]
] |
2506.02652 | Jos\'e C\'aceres | Jos\'e C\'aceres and Ignacio M. Pelayo | Not every graph can be reconstructed from its boundary distance matrix | arXiv admin note: text overlap with arXiv:2404.04039 | null | null | null | math.CO | http://creativecommons.org/licenses/by-nc-sa/4.0/ | A vertex $v$ of a connected graph $G$ is said to be a boundary vertex of $G$ if for some other vertex $u$ of $G$, no neighbor of $v$ is further away from $u$ than $v$. The boundary $\partial(G)$ of $G$ is the set of all of its boundary vertices.
The boundary distance matrix $\hat{D}_G$ of a graph $G=([n],E)$ is the square matrix of order $\kappa$, being $\kappa$ the order of $\partial(G)$, such that for every $i,j\in \partial(G)$, $[\hat{D}_G]_{ij}=d_G(i,j)$.
In a recent paper [doi.org/10.7151/dmgt.2567], it was shown that if a graph $G$ is either a block graph or a unicyclic graph, then $G$ is uniquely determined by the boundary distance matrix $\hat{D}_{G}$ of $G$, and it was also conjectured that this statement holds for every connected graph $G$, whenever both the order $n$ and the boundary (and thus also the boundary distance matrix) of $G$ are prefixed.
After proving that this conjecture is true for several graph families, such as being of diameter 2, having order at most $n=6$ or being Ptolemaic, we show that this statement does not hold when considering, for example, either the family of split graphs of diameter 3 and order at least $n=10$ or the family of distance-hereditary graphs of order at least $n=8$. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:04:04 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Cáceres",
"José",
""
],
[
"Pelayo",
"Ignacio M.",
""
]
] |
2506.02653 | Balendra Pratap Singh | Balendra Pratap Singh and Md Sabir Ali | Thermodynamics and Shadows of Kerr black holes endowed with a global monopole charge | 25 pages, 13 figures, four tables. Comments are welcome | null | null | null | gr-qc | http://creativecommons.org/licenses/by-nc-sa/4.0/ | In this article, we present the thermodynamic and shadow properties of rotating black holes with global monopole charge. The angular diameter of Sgr A$^{*}$ black hole is 48.7 $\pm$ 7 $\mu as$, which is 8 $kps$ far away having a mass of $M = 4.0_{-0.6}^{+1.1} \times 10^6 M\odot$ as observed by Event Horizon Telescope and for the M87 black hole, the observed angular diameter is $\theta_d = 42 \pm 3 \mu$as, which is almost $16$ $Mpc$ far away with a mass of $M = (6.5 \pm 0.7) \times 10^9 M_\odot$. The global monopole charge parameter $\alpha$ strongly affects the shape and size of the black hole shadow. We derived all the necessary equations to obtain the angular diameter of the rotating black hole shadow with the effect of the global monopole charge parameter $\alpha$. For $\alpha$
$\in$ (0, 0.08) with $a$ $\in$ $(0.7 M, 0.99 M)$, the angular diameter of M87 black hole shadow varies from $39$ $\mu as$ to $51$ $\mu as$. The angular diameter of Sgr A$^{*}$ black hole with global monopole charge parameter $\alpha$ $\in$ (0, 0.04) and $a$ $\in$ $(0.7 M, 0.99 M)$, varies from $50$ $\mu as$ to $55$ $\mu as$. For bound values of $\alpha$ and $a$, our results are consistent with the EHT observations. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:06:31 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Singh",
"Balendra Pratap",
""
],
[
"Ali",
"Md Sabir",
""
]
] |
2506.02654 | Shiyu Shen | Shiyu Shen, Bin Pan, Guirong Xue | A Pretrained Probabilistic Transformer for City-Scale Traffic Volume Prediction | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | City-scale traffic volume prediction plays a pivotal role in intelligent transportation systems, yet remains a challenge due to the inherent incompleteness and bias in observational data. Although deep learning-based methods have shown considerable promise, most existing approaches produce deterministic point estimates, thereby neglecting the uncertainty arising from unobserved traffic flows. Furthermore, current models are typically trained in a city-specific manner, which hinders their generalizability and limits scalability across diverse urban contexts. To overcome these limitations, we introduce TrafficPPT, a Pretrained Probabilistic Transformer designed to model traffic volume as a distributional aggregation of trajectories. Our framework fuses heterogeneous data sources-including real-time observations, historical trajectory data, and road network topology-enabling robust and uncertainty-aware traffic inference. TrafficPPT is initially pretrained on large-scale simulated data spanning multiple urban scenarios, and later fine-tuned on target cities to ensure effective domain adaptation. Experiments on real-world datasets show that TrafficPPT consistently surpasses state-of-the-art baselines, particularly under conditions of extreme data sparsity. Code will be open. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:07:29 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Shen",
"Shiyu",
""
],
[
"Pan",
"Bin",
""
],
[
"Xue",
"Guirong",
""
]
] |
2506.02655 | Kaito Fujii | Kaito Fujii | The power of mediators: Price of anarchy and stability in Bayesian games with submodular social welfare | null | null | null | null | cs.GT cs.DS | http://creativecommons.org/licenses/by/4.0/ | This paper investigates the role of mediators in Bayesian games by examining their impact on social welfare through the price of anarchy (PoA) and price of stability (PoS). Mediators can communicate with players to guide them toward equilibria of varying quality, and different communication protocols lead to a variety of equilibrium concepts collectively known as Bayes (coarse) correlated equilibria. To analyze these equilibrium concepts, we consider a general class of Bayesian games with submodular social welfare, which naturally extends valid utility games and their variant, basic utility games. These frameworks, introduced by Vetta (2002), have been developed to analyze the social welfare guarantees of equilibria in games such as competitive facility location, influence maximization, and other resource allocation problems.
We provide upper and lower bounds on the PoA and PoS for a broad class of Bayes (coarse) correlated equilibria. Central to our analysis is the strategy representability gap, which measures the multiplicative gap between the optimal social welfare achievable with and without knowledge of other players' types. For monotone submodular social welfare functions, we show that this gap is $1-1/\mathrm{e}$ for independent priors and $\Theta(1/\sqrt{n})$ for correlated priors, where $n$ is the number of players. These bounds directly lead to upper and lower bounds on the PoA and PoS for various equilibrium concepts, while we also derive improved bounds for specific concepts by developing smoothness arguments. Notably, we identify a fundamental gap in the PoA and PoS across different classes of Bayes correlated equilibria, highlighting essential distinctions among these concepts. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:08:03 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Fujii",
"Kaito",
""
]
] |
2506.02656 | Zeshan Haider | Zeshan Haider | The Initial Experimental Guide to Implement QKD Polarization Encoding System at Telecom Wavelength | null | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The experimental implementation of the polarization encoding system is presented using weak coherent pulses at a low enough frequency. The optical pulses are generated through intensity modulation at the repetition rate of $10$ pulses/sec with the intensity of $7\mu W$, whose polarization is modulated by the Pockel cell. For a $ 5$ km long fiber link, the key rate is thus limited to $ 10$ bits/sec accompanied by temporal synchronization among the Intensity Modulator (IM), Pockel cell, and the gate of the detector. The system was continuously operated for $5$ minutes, and $3000$ binary bits were distributed between two nodes with an error rate less than unity. We have used only a single polarization basis, i.e., horizontal and vertical polarizations, as the objective of this experiment is to validate the feasibility of the setup under simplified conditions, intended primarily for the laboratory-scale demonstrations. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:09:31 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Haider",
"Zeshan",
""
]
] |
2506.02657 | Trinh Van Chien | Tam Ninh Thi-Thanh, Trinh Van Chien, Hung Tran, Nguyen Hoai Son, Van Nhan Vo | Maximizing the Promptness of Metaverse Systems using Edge Computing by Deep Reinforcement Learning | 6 pages, 3 figures, and 2 tables. Published by IEEE at ATC2024 | null | null | null | cs.IT cs.LG math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Metaverse and Digital Twin (DT) have attracted much academic and industrial attraction to approach the future digital world. This paper introduces the advantages of deep reinforcement learning (DRL) in assisting Metaverse system-based Digital Twin. In this system, we assume that it includes several Metaverse User devices collecting data from the real world to transfer it into the virtual world, a Metaverse Virtual Access Point (MVAP) undertaking the processing of data, and an edge computing server that receives the offloading data from the MVAP. The proposed model works under a dynamic environment with various parameters changing over time. The experiment results show that our proposed DRL algorithm is suitable for offloading tasks to ensure the promptness of DT in a dynamic environment. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:10:31 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Thi-Thanh",
"Tam Ninh",
""
],
[
"Van Chien",
"Trinh",
""
],
[
"Tran",
"Hung",
""
],
[
"Son",
"Nguyen Hoai",
""
],
[
"Vo",
"Van Nhan",
""
]
] |
2506.02659 | Manon Reusens | Manon Reusens, Bart Baesens, David Jurgens | Are Economists Always More Introverted? Analyzing Consistency in Persona-Assigned LLMs | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Personalized Large Language Models (LLMs) are increasingly used in diverse applications, where they are assigned a specific persona - such as a happy high school teacher - to guide their responses. While prior research has examined how well LLMs adhere to predefined personas in writing style, a comprehensive analysis of consistency across different personas and task types is lacking. In this paper, we introduce a new standardized framework to analyze consistency in persona-assigned LLMs. We define consistency as the extent to which a model maintains coherent responses when assigned the same persona across different tasks and runs. Our framework evaluates personas across four different categories (happiness, occupation, personality, and political stance) spanning multiple task dimensions (survey writing, essay generation, social media post generation, single turn, and multi-turn conversations). Our findings reveal that consistency is influenced by multiple factors, including the assigned persona, stereotypes, and model design choices. Consistency also varies across tasks, increasing with more structured tasks and additional context. All code is available on GitHub. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:12:23 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Reusens",
"Manon",
""
],
[
"Baesens",
"Bart",
""
],
[
"Jurgens",
"David",
""
]
] |
2506.02660 | Gabriele Digregorio | Gabriele Digregorio, Salvatore Maccarrone, Mario D'Onghia, Luigi Gallo, Michele Carminati, Mario Polino, Stefano Zanero | Tarallo: Evading Behavioral Malware Detectors in the Problem Space | null | International Conference on Detection of Intrusions and Malware, and Vulnerability Assessment, 2024 | 10.1007/978-3-031-64171-8_7 | null | cs.CR | http://creativecommons.org/licenses/by/4.0/ | Machine learning algorithms can effectively classify malware through dynamic behavior but are susceptible to adversarial attacks. Existing attacks, however, often fail to find an effective solution in both the feature and problem spaces. This issue arises from not addressing the intrinsic nondeterministic nature of malware, namely executing the same sample multiple times may yield significantly different behaviors. Hence, the perturbations computed for a specific behavior may be ineffective for others observed in subsequent executions. In this paper, we show how an attacker can augment their chance of success by leveraging a new and more efficient feature space algorithm for sequential data, which we have named PS-FGSM, and by adopting two problem space strategies specially tailored to address nondeterminism in the problem space. We implement our novel algorithm and attack strategies in Tarallo, an end-to-end adversarial framework that significantly outperforms previous works in both white and black-box scenarios. Our preliminary analysis in a sandboxed environment and against two RNN-based malware detectors, shows that Tarallo achieves a success rate up to 99% on both feature and problem space attacks while significantly minimizing the number of modifications required for misclassification. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:12:43 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Digregorio",
"Gabriele",
""
],
[
"Maccarrone",
"Salvatore",
""
],
[
"D'Onghia",
"Mario",
""
],
[
"Gallo",
"Luigi",
""
],
[
"Carminati",
"Michele",
""
],
[
"Polino",
"Mario",
""
],
[
"Zanero",
"Stefano",
""
]
] |
2506.02661 | Mingyang Huang | Mingyang Huang, Peng Zhang, Bang Zhang | MotionRAG-Diff: A Retrieval-Augmented Diffusion Framework for Long-Term Music-to-Dance Generation | 12 pages, 5 figures | null | null | null | cs.SD cs.CV cs.GR eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Generating long-term, coherent, and realistic music-conditioned dance sequences remains a challenging task in human motion synthesis. Existing approaches exhibit critical limitations: motion graph methods rely on fixed template libraries, restricting creative generation; diffusion models, while capable of producing novel motions, often lack temporal coherence and musical alignment. To address these challenges, we propose $\textbf{MotionRAG-Diff}$, a hybrid framework that integrates Retrieval-Augmented Generation (RAG) with diffusion-based refinement to enable high-quality, musically coherent dance generation for arbitrary long-term music inputs. Our method introduces three core innovations: (1) A cross-modal contrastive learning architecture that aligns heterogeneous music and dance representations in a shared latent space, establishing unsupervised semantic correspondence without paired data; (2) An optimized motion graph system for efficient retrieval and seamless concatenation of motion segments, ensuring realism and temporal coherence across long sequences; (3) A multi-condition diffusion model that jointly conditions on raw music signals and contrastive features to enhance motion quality and global synchronization. Extensive experiments demonstrate that MotionRAG-Diff achieves state-of-the-art performance in motion quality, diversity, and music-motion synchronization accuracy. This work establishes a new paradigm for music-driven dance generation by synergizing retrieval-based template fidelity with diffusion-based creative enhancement. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:12:48 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Huang",
"Mingyang",
""
],
[
"Zhang",
"Peng",
""
],
[
"Zhang",
"Bang",
""
]
] |
2506.02662 | Yuchen Bao | Yuchen Bao, Zhenwei Li, Hongwei Ge, Xuefei Chen, Zhanwen Han | Be star + He star binary as an indicator of binary mass transfer phase | accepted for publication in ApJ | null | null | null | astro-ph.SR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The rapid rotation of Be stars is supposed to mainly originate from binary evolution. In recent years, more and more Be stars with helium (He) star companions have been discovered, which provides a significant opportunity to study binary interaction physics. In this work, we perform binary population synthesis with an updated binary mass transfer stability criterion and try to understand the details of mass transfer processes by constructing a series of Be star + He star (BeHe) binary populations. We found that the simulations and the observations can be divided into two groups according to the masses of components, corresponding to the two distinct evolutionary processes during the mass transfer. In particular, we found that the mass ratios of BeHe binaries may be taken as a probe of the initial mass ratios of the primordial binaries. Moreover, the results suggest that a higher mass transfer efficiency ($\gtrsim 0.5$) supports the observations better. The simulations predicted too many Be star binaries experiencing Case B mass transfer, which conflicts with the observations. The reason is due to either observational selection effects or unclear physical factors. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:13:00 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Bao",
"Yuchen",
""
],
[
"Li",
"Zhenwei",
""
],
[
"Ge",
"Hongwei",
""
],
[
"Chen",
"Xuefei",
""
],
[
"Han",
"Zhanwen",
""
]
] |
2506.02663 | Shubo Zhao | Shubo Zhao and Qinghai Zhang | Fourth-order Adaptive Mesh Refinement both in space and in time for incompressible Navier-Stokes equations with Dirichlet boundary conditions | null | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by/4.0/ | We present a fourth-order projection method with adaptive mesh refinement (AMR) for numerically solving the incompressible Navier-Stokes equations (INSE) with subcycling in time. Our method features (i) a reformulation of INSE so that the velocity divergence decays exponentially on the coarsest level, (ii) a derivation of coarse-fine interface conditions that preserves the decay of velocity divergence on any refinement level of the AMR hierarchy, (iii) an approximation of the coarse-fine interface conditions via spatiotemporal interpolations to facilitate subcycling in time, (iv) enforcing to machine precision solvability conditions of elliptic equations over each connected component of the subdomain covered by any refinement level, (v) a composite projection for synchronizing multiple levels, and (vi) geometric multigrid for solving linear systems with optimal complexity. Different from current block-structured AMR algorithms, our method never adopts refluxing at the coarse-fine interface, nor is fine-to-coarse averaging applied to projected velocities. Results of numerical tests demonstrate the high accuracy and efficiency of the proposed method. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:14:13 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Zhao",
"Shubo",
""
],
[
"Zhang",
"Qinghai",
""
]
] |
2506.02664 | Hugo Tabanelli | Hugo Tabanelli, Pierre Mergny, Lenka Zdeborova, Florent Krzakala | Computational Thresholds in Multi-Modal Learning via the Spiked Matrix-Tensor Model | null | null | null | null | stat.ML cond-mat.dis-nn cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the recovery of multiple high-dimensional signals from two noisy, correlated modalities: a spiked matrix and a spiked tensor sharing a common low-rank structure. This setting generalizes classical spiked matrix and tensor models, unveiling intricate interactions between inference channels and surprising algorithmic behaviors. Notably, while the spiked tensor model is typically intractable at low signal-to-noise ratios, its correlation with the matrix enables efficient recovery via Bayesian Approximate Message Passing, inducing staircase-like phase transitions reminiscent of neural network phenomena. In contrast, empirical risk minimization for joint learning fails: the tensor component obstructs effective matrix recovery, and joint optimization significantly degrades performance, highlighting the limitations of naive multi-modal learning. We show that a simple Sequential Curriculum Learning strategy-first recovering the matrix, then leveraging it to guide tensor recovery-resolves this bottleneck and achieves optimal weak recovery thresholds. This strategy, implementable with spectral methods, emphasizes the critical role of structural correlation and learning order in multi-modal high-dimensional inference. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:14:34 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Tabanelli",
"Hugo",
""
],
[
"Mergny",
"Pierre",
""
],
[
"Zdeborova",
"Lenka",
""
],
[
"Krzakala",
"Florent",
""
]
] |
2506.02665 | Tianci Liu | Tianci Liu and Tong Yang and Quan Zhang and Qi Lei | Beyond Invisibility: Learning Robust Visible Watermarks for Stronger Copyright Protection | UAI 2025 | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As AI advances, copyrighted content faces growing risk of unauthorized use, whether through model training or direct misuse. Building upon invisible adversarial perturbation, recent works developed copyright protections against specific AI techniques such as unauthorized personalization through DreamBooth that are misused. However, these methods offer only short-term security, as they require retraining whenever the underlying model architectures change. To establish long-term protection aiming at better robustness, we go beyond invisible perturbation, and propose a universal approach that embeds \textit{visible} watermarks that are \textit{hard-to-remove} into images. Grounded in a new probabilistic and inverse problem-based formulation, our framework maximizes the discrepancy between the \textit{optimal} reconstruction and the original content. We develop an effective and efficient approximation algorithm to circumvent a intractable bi-level optimization. Experimental results demonstrate superiority of our approach across diverse scenarios. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:14:46 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Liu",
"Tianci",
""
],
[
"Yang",
"Tong",
""
],
[
"Zhang",
"Quan",
""
],
[
"Lei",
"Qi",
""
]
] |
2506.02666 | Nikolaos Miridakis | Nikolaos I. Miridakis, Panagiotis A. Karkazis | Spatially Correlated multi-RIS Communication: The Effect of Inter-Operator Interference | Submitted to IEEE Journal. arXiv admin note: text overlap with arXiv:2403.00349 | null | null | null | cs.IT cs.PF math.IT | http://creativecommons.org/licenses/by/4.0/ | A multi-operator wireless communication system is studied where each operator is equipped with a reconfigurable intelligent surface (RIS) to enhance its communication quality. RISs controlled by different operators affect the system performance of one another due to the inherently rapid phase shift adjustments that occur on an independent basis. The system performance of such a communication scenario is analytically studied for the practical case where spatial correlation occurs at RIS of arbitrary size. The proposed framework is quite general since it is analyzed under Nakagami-$m$ channel fading conditions. Finally, the derived analytical results are verified via numerical and simulation trials as well as some new and useful engineering outcomes are revealed. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:14:53 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Miridakis",
"Nikolaos I.",
""
],
[
"Karkazis",
"Panagiotis A.",
""
]
] |
2506.02668 | Frederico Metelo | Frederico Metelo, Alexandre Oliveira, Stevo Rackovi\'c, Pedro \'Akos Costa, Cl\'audia Soares | FAuNO: Semi-Asynchronous Federated Reinforcement Learning Framework for Task Offloading in Edge Systems | null | null | null | null | cs.AI cs.LG | http://creativecommons.org/licenses/by-sa/4.0/ | Edge computing addresses the growing data demands of connected-device networks by placing computational resources closer to end users through decentralized infrastructures. This decentralization challenges traditional, fully centralized orchestration, which suffers from latency and resource bottlenecks. We present \textbf{FAuNO} -- \emph{Federated Asynchronous Network Orchestrator} -- a buffered, asynchronous \emph{federated reinforcement-learning} (FRL) framework for decentralized task offloading in edge systems. FAuNO adopts an actor-critic architecture in which local actors learn node-specific dynamics and peer interactions, while a federated critic aggregates experience across agents to encourage efficient cooperation and improve overall system performance. Experiments in the \emph{PeersimGym} environment show that FAuNO consistently matches or exceeds heuristic and federated multi-agent RL baselines in reducing task loss and latency, underscoring its adaptability to dynamic edge-computing scenarios. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:15:03 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Metelo",
"Frederico",
""
],
[
"Oliveira",
"Alexandre",
""
],
[
"Racković",
"Stevo",
""
],
[
"Costa",
"Pedro Ákos",
""
],
[
"Soares",
"Cláudia",
""
]
] |
2506.02669 | Yerali Gandica | David Hernandez, Guillaume Deffuant and Yerali Gandica | Bounded confidence dynamics generates opinion cascades on a growing scale-free network | 5 figures, 12 pages | null | null | null | physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the pairwise bounded confidence model on scale-free networks where new agents regularly arrive over time. The probability that arriving agents form links to preexisting ones depends on both agent degree and opinion proximity. In parameter value ranges where both factors impact the link choice, a new phenomenon is observed. Minor clusters continuously form on the periphery of the opinion space and remain stable for a time, before suddenly merging with a major cluster in the network. We label these processes as "opinion cascades", and analyse their origin and behavior. They are triggered by the arrival of agents acting as "bridges" between the previously disconnected minor and major clusters. Lastly, we propose theoretical approximations to describe the varying shapes and merging behavior of opinion cascades under different conditions. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:15:58 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Hernandez",
"David",
""
],
[
"Deffuant",
"Guillaume",
""
],
[
"Gandica",
"Yerali",
""
]
] |
2506.02670 | Benjamin Meco | Stig Lundgren, Benjamin Meco | A generalization of the ADM mass for asymptotically Euclidean manifolds of weak regularity | 38 pages | null | null | null | math.DG gr-qc | http://creativecommons.org/licenses/by/4.0/ | We propose a new definition of the ADM mass for asymptotically Euclidean manifolds inspired by the definition of mass for weakly regular asymptotically hyperbolic manifolds by Gicquaud and Sakovich. This version of the mass allows one to work with metrics of local Sobolev regularity $ W^{1,2}_\text{loc} \cap L^\infty $ and we show, under suitable asymptotic assumptions, that the mass is finite, invariant under a change of coordinates at infinity and that it agrees with the classical ADM mass in the smooth setting. We also provide an expression in terms of the Ricci tensor that agrees with the Ricci version of the ADM mass studied by Herzlich. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:16:05 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Lundgren",
"Stig",
""
],
[
"Meco",
"Benjamin",
""
]
] |
2506.02671 | Jiazhen Huang | Xiao Chen, Jiazhen Huang, Qinting Jiang, Fanding Huang, Xianghua Fu, Jingyan Jiang, Zhi Wang | Small Aid, Big Leap: Efficient Test-Time Adaptation for Vision-Language Models with AdaptNet | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Test-time adaptation (TTA) has emerged as a critical technique for enhancing the generalization capability of vision-language models (VLMs) during inference. However, existing approaches often incur substantial computational costs and exhibit poor scalability, primarily due to sample-wise adaptation granularity and reliance on costly auxiliary designs such as data augmentation. To address these limitations, we introduce SAIL (Small Aid, Big Leap), a novel adapter-based TTA framework that leverages a lightweight, learnable AdaptNet to enable efficient and scalable model adaptation. As SAIL's core, a frozen pre-trained VLM collaborates with AdaptNet through a confidence-based interpolation weight, generating robust predictions during inference. These predictions serve as self-supervised targets to align AdaptNet's outputs through efficient batch-wise processing, dramatically reducing computational costs without modifying the VLM or requiring memory caches. To mitigate catastrophic forgetting during continual adaptation, we propose a gradient-aware reset strategy driven by a gradient drift indicator (GDI), which dynamically detects domain transitions and strategically resets AdaptNet for stable adaptation. Extensive experiments across diverse benchmarks on two scenarios demonstrate that SAIL achieves state-of-the-art performance while maintaining low computational costs. These results highlight SAIL's effectiveness, efficiency and scalability for real-world deployment. The code will be released upon acceptance. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:16:51 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Chen",
"Xiao",
""
],
[
"Huang",
"Jiazhen",
""
],
[
"Jiang",
"Qinting",
""
],
[
"Huang",
"Fanding",
""
],
[
"Fu",
"Xianghua",
""
],
[
"Jiang",
"Jingyan",
""
],
[
"Wang",
"Zhi",
""
]
] |
2506.02673 | Martijn Kemerink | H. Mager, A. A. Butkevich, S. Klubertz, S. V. Haridas, O. Shyshov, V. C. Wakchaure, J. Borstelmann, I. Michalsky, M. Garc\'ia-Iglesias, V. Rodr\'iguez, D. Gonz\'alez-Rodr\'iguez, A. R. A. Palmans, M. Kivala, M. von Delius and M. Kemerink | Long-range electrical conductivity in non-$\pi$-conjugated organic molecular materials | null | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Electronic conductivity in organic materials is well-established. Moreover, semiconductive or metallic behavior in (quasi) 0-, 1-, 2- and 3-dimensional carbon-based materials has been demonstrated and understood and is nowadays widely applied in commercial devices. Despite the large structural variety, these materials commonly have an extended $\pi$-system, formed by sp2-hybridized carbon atoms, which is responsible for the conductivity. Here, we present a class of organic molecular materials that, despite the absence of an extended $\pi$-system, show a distinct direct current conductivity in quasi-1D supramolecular stacks of small organic molecules. Long-range conductivity takes place by removal of an electron from the highest occupied molecular orbital, i.e. oxidation, followed by charge transfer between neighboring molecules. In thin-film devices, the resulting transport band becomes energetically accessible to charges from the contacts by interfacial dipoles stemming from a dipolar group in the molecule. Long-range order in the form of the supramolecular polymers with lengths exceeding several micrometers enhances the macroscopic conductivity but is not essential. The results presented herein show a new way to incorporate electronic functionality in organic materials and present a new class of organic conductors that lack the extended $\pi$-electron-system of conventional organic semiconductors. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:19:04 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Mager",
"H.",
""
],
[
"Butkevich",
"A. A.",
""
],
[
"Klubertz",
"S.",
""
],
[
"Haridas",
"S. V.",
""
],
[
"Shyshov",
"O.",
""
],
[
"Wakchaure",
"V. C.",
""
],
[
"Borstelmann",
"J.",
""
],
[
"Michalsky",
"I.",
""
],
[
"García-Iglesias",
"M.",
""
],
[
"Rodríguez",
"V.",
""
],
[
"González-Rodríguez",
"D.",
""
],
[
"Palmans",
"A. R. A.",
""
],
[
"Kivala",
"M.",
""
],
[
"von Delius",
"M.",
""
],
[
"Kemerink",
"M.",
""
]
] |
2506.02674 | Shipeng Ye | Lingsheng Chen, Shipeng Ye, Xiaoqi Li | Decentralized COVID-19 Health System Leveraging Blockchain | 21 pages, 5 figures | null | null | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the development of the Internet, the amount of data generated by the medical industry each year has grown exponentially. The Electronic Health Record (EHR) manages the electronic data generated during the user's treatment process. Typically, an EHR data manager belongs to a medical institution. This traditional centralized data management model has many unreasonable or inconvenient aspects, such as difficulties in data sharing, and it is hard to verify the authenticity and integrity of the data. The decentralized, non-forgeable, data unalterable and traceable features of blockchain are in line with the application requirements of EHR. This paper takes the most common COVID-19 as the application scenario and designs a COVID-19 health system based on blockchain, which has extensive research and application value. Considering that the public and transparent nature of blockchain violates the privacy requirements of some health data, in the system design stage, from the perspective of practical application, the data is divided into public data and private data according to its characteristics. For private data, data encryption methods are adopted to ensure data privacy. The searchable encryption technology is combined with blockchain technology to achieve the retrieval function of encrypted data. Then, the proxy re-encryption technology is used to realize authorized access to data. In the system implementation part, based on the Hyperledger Fabric architecture, some functions of the system design are realized, including data upload, retrieval of the latest data and historical data. According to the environment provided by the development architecture, Go language chaincode (smart contract) is written to implement the relevant system functions. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:19:47 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Chen",
"Lingsheng",
""
],
[
"Ye",
"Shipeng",
""
],
[
"Li",
"Xiaoqi",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.