id
string | submitter
string | authors
string | title
string | comments
string | journal-ref
string | doi
string | report-no
string | categories
string | license
string | abstract
string | versions
list | update_date
timestamp[s] | authors_parsed
sequence |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2505.22774 | Kaja Dobrovoljc | Kaja Dobrovoljc | Counting trees: A treebank-driven exploration of syntactic variation in speech and writing across languages | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | This paper presents a novel treebank-driven approach to comparing syntactic structures in speech and writing using dependency-parsed corpora. Adopting a fully inductive, bottom-up method, we define syntactic structures as delexicalized dependency (sub)trees and extract them from spoken and written Universal Dependencies (UD) treebanks in two syntactically distinct languages, English and Slovenian. For each corpus, we analyze the size, diversity, and distribution of syntactic inventories, their overlap across modalities, and the structures most characteristic of speech. Results show that, across both languages, spoken corpora contain fewer and less diverse syntactic structures than their written counterparts, with consistent cross-linguistic preferences for certain structural types across modalities. Strikingly, the overlap between spoken and written syntactic inventories is very limited: most structures attested in speech do not occur in writing, pointing to modality-specific preferences in syntactic organization that reflect the distinct demands of real-time interaction and elaborated writing. This contrast is further supported by a keyness analysis of the most frequent speech-specific structures, which highlights patterns associated with interactivity, context-grounding, and economy of expression. We argue that this scalable, language-independent framework offers a useful general method for systematically studying syntactic variation across corpora, laying the groundwork for more comprehensive data-driven theories of grammar in use. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:43:26 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Dobrovoljc",
"Kaja",
""
]
] |
2505.22775 | Abhik Bhattacharjee | Arka Banerjee, Abhik Bhattacharjee, Subinoy Das, Anshuman Maharana, Ravi Kumar Sharma | Forecasting Constraints on Non-Thermal Light Massive Relics from Future CMB Experiments | 13 pages text + 2 pages appendices + 2 pages references, 12 figures. Comments are welcome | null | null | null | astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | In this work, we present Fisher forecasts on non-thermal LiMR models for the upcoming CMB Stage IV experiment -- particularly focusing on a model of inflaton/moduli decay giving rise to non-thermally distributed dark sector particles, and also comparing our results with those for sterile particles following the Dodelson-Widrow distribution. Two independent parameters, the effective number of extra relativistic species $\Delta N_\mathrm{eff}$ and the effective mass $M_\mathrm{sp}^\mathrm{eff}$ of the relic, influence linear cosmological observables. We find $\Delta N_\mathrm{eff}$ to be more tightly constrained with $\sigma(\Delta N_\mathrm{eff})\sim10^{-3}$, for a less abundant, heavier LiMR which becomes fully non-relativistic around matter-radiation equality than a more abundant, lighter LiMR which becomes fully non-relativistic just after recombination, for which $\sigma(\Delta N_\mathrm{eff})\sim10^{-2}$. The uncertainties on $M_\mathrm{sp}^\mathrm{eff}$ differ by a factor of $\sim3$ between the two cases. Our analysis also reveals distinct parameter correlations: the phenomenological parameters $\{\Delta N_\mathrm{eff}, M_\mathrm{sp}^\mathrm{eff}\}$ are found to be negatively correlated for the former case and positively correlated for the latter. We obtain similar constraints on the cosmological parameters (in either case) for both the inflaton/moduli decay and the Dodelson-Widrow models when the first two moments of the LiMR distribution function, related to the phenomenological parameters, are matched. Finally, by constructing a modified distribution that matches the first two moments of the Dodelson-Widrow but deviates maximally in the third moment, we demonstrate that CMB Stage IV data is not expected to be sensitive to higher moments of the distribution. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:45:11 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Banerjee",
"Arka",
""
],
[
"Bhattacharjee",
"Abhik",
""
],
[
"Das",
"Subinoy",
""
],
[
"Maharana",
"Anshuman",
""
],
[
"Sharma",
"Ravi Kumar",
""
]
] |
2505.22776 | Tren M.J.T. Baltussen | Merlijne Geurts, Tren Baltussen, Alexander Katriniok, Maurice Heemels | A Contingency Model Predictive Control Framework for Safe Learning | 6 pages, 3 figures. To appear in the IEEE Control Systems Letters | null | 10.1109/LCSYS.2025.3575191 | null | math.OC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This research introduces a multi-horizon contingency model predictive control (CMPC) framework in which classes of robust MPC (RMPC) algorithms are combined with classes of learning-based MPC (LB-MPC) algorithms to enable safe learning. We prove that the CMPC framework inherits the robust recursive feasibility properties of the underlying RMPC scheme, thereby ensuring safety of the CMPC in the sense of constraint satisfaction. The CMPC leverages the LB-MPC to safely learn the unmodeled dynamics to reduce conservatism and improve performance compared to standalone RMPC schemes, which are conservative in nature. In addition, we present an implementation of the CMPC framework that combines a particular RMPC and a Gaussian Process MPC scheme. A simulation study on automated lane merging demonstrates the advantages of our general CMPC framework. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:45:23 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Geurts",
"Merlijne",
""
],
[
"Baltussen",
"Tren",
""
],
[
"Katriniok",
"Alexander",
""
],
[
"Heemels",
"Maurice",
""
]
] |
2505.22777 | John Mendonca | John Mendon\c{c}a, Alon Lavie, Isabel Trancoso | MEDAL: A Framework for Benchmarking LLMs as Multilingual Open-Domain Chatbots and Dialogue Evaluators | May ARR | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | As the capabilities of chatbots and their underlying LLMs continue to dramatically improve, evaluating their performance has increasingly become a major blocker to their further development. A major challenge is the available benchmarking datasets, which are largely static, outdated, and lacking in multilingual coverage, limiting their ability to capture subtle linguistic and cultural variations. This paper introduces MEDAL, an automated multi-agent framework for generating, evaluating, and curating more representative and diverse open-domain dialogue evaluation benchmarks. Our approach leverages several state-of-the-art LLMs to generate user-chatbot multilingual dialogues, conditioned on varied seed contexts. A strong LLM (GPT-4.1) is then used for a multidimensional analysis of the performance of the chatbots, uncovering noticeable cross-lingual performance differences. Guided by this large-scale evaluation, we curate a new meta-evaluation multilingual benchmark and human-annotate samples with nuanced quality judgments. This benchmark is then used to assess the ability of several reasoning and non-reasoning LLMs to act as evaluators of open-domain dialogues. We find that current LLMs struggle to detect nuanced issues, particularly those involving empathy and reasoning. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:45:42 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Mendonça",
"John",
""
],
[
"Lavie",
"Alon",
""
],
[
"Trancoso",
"Isabel",
""
]
] |
2505.22778 | Sarah Meiklejohn | Sarah Meiklejohn, Hayden Blauzvern, Mihai Maruseac, Spencer Schrock, Laurent Simon, Ilia Shumailov | Machine Learning Models Have a Supply Chain Problem | 12 pages, four figures and one table | ICML 2025 | null | null | cs.LG cs.CR | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Powerful machine learning (ML) models are now readily available online, which creates exciting possibilities for users who lack the deep technical expertise or substantial computing resources needed to develop them. On the other hand, this type of open ecosystem comes with many risks. In this paper, we argue that the current ecosystem for open ML models contains significant supply-chain risks, some of which have been exploited already in real attacks. These include an attacker replacing a model with something malicious (e.g., malware), or a model being trained using a vulnerable version of a framework or on restricted or poisoned data. We then explore how Sigstore, a solution designed to bring transparency to open-source software supply chains, can be used to bring transparency to open ML models, in terms of enabling model publishers to sign their models and prove properties about the datasets they use. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:47:14 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Meiklejohn",
"Sarah",
""
],
[
"Blauzvern",
"Hayden",
""
],
[
"Maruseac",
"Mihai",
""
],
[
"Schrock",
"Spencer",
""
],
[
"Simon",
"Laurent",
""
],
[
"Shumailov",
"Ilia",
""
]
] |
2505.22779 | Mohammad Helal Uddin | Mohammad Helal Uddin and Sabur Baidya | Predicting Human Depression with Hybrid Data Acquisition utilizing Physical Activity Sensing and Social Media Feeds | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Mental disorders including depression, anxiety, and other neurological disorders pose a significant global challenge, particularly among individuals exhibiting social avoidance tendencies. This study proposes a hybrid approach by leveraging smartphone sensor data measuring daily physical activities and analyzing their social media (Twitter) interactions for evaluating an individual's depression level. Using CNN-based deep learning models and Naive Bayes classification, we identify human physical activities accurately and also classify the user sentiments. A total of 33 participants were recruited for data acquisition, and nine relevant features were extracted from the physical activities and analyzed with their weekly depression scores, evaluated using the Geriatric Depression Scale (GDS) questionnaire. Of the nine features, six are derived from physical activities, achieving an activity recognition accuracy of 95%, while three features stem from sentiment analysis of Twitter activities, yielding a sentiment analysis accuracy of 95.6%. Notably, several physical activity features exhibited significant correlations with the severity of depression symptoms. For classifying the depression severity, a support vector machine (SVM)-based algorithm is employed that demonstrated a very high accuracy of 94%, outperforming alternative models, e.g., the multilayer perceptron (MLP) and k-nearest neighbor. It is a simple approach yet highly effective in the long run for monitoring depression without breaching personal privacy. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:47:34 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Uddin",
"Mohammad Helal",
""
],
[
"Baidya",
"Sabur",
""
]
] |
2505.22780 | Hammed Oyekan | Hammed Oyekan | Depth to magnetic source estimation using TDX contour | null | null | null | null | physics.geo-ph | http://creativecommons.org/licenses/by/4.0/ | Accurate depth estimation of magnetic sources plays a crucial role in various geophysical applications, including mineral exploration, resource assessments, regional hydrocarbon exploration, and geological mapping. Thus, this abstract presents a fast and simple method of estimating the depth of a magnetic body using the TDX derivative of the total magnetic field. TDX is a first-order derivative of the magnetic field that, in addition to edge detection, is less affected by noise, allowing for better depth resolution. The reduced sensitivity to noise enables a clearer estimation of depth and enhances the accuracy of the depth determination process. The TDX, as a variant of the phase derivative, is independent of magnetization and can be used to identify the edge of a magnetic body. In addition to excelling at edge detection, they can also estimate the depth of the magnetic source producing the anomalies. In this study, we explore the utilization of contour of the TDX derivative for estimating depth, assuming a vertical contact source. We demonstrate the effectiveness of the method using a two-prism block model and a simple bishop model with a uniform susceptibility of 0.001 cgs. The results agree with the known depth, providing evidence of the reliability of the method despite the restrictive nature of the assumption, especially for the Bishop model, where there are numerous fault structures. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:47:56 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Oyekan",
"Hammed",
""
]
] |
2505.22781 | Antonio Ocello | Antonio Ocello and Daniil Tiapkin and Lorenzo Mancini and Mathieu Lauri\`ere and Eric Moulines | Finite-Sample Convergence Bounds for Trust Region Policy Optimization in Mean-Field Games | null | null | null | null | stat.ML cs.LG math.ST stat.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce Mean-Field Trust Region Policy Optimization (MF-TRPO), a novel algorithm designed to compute approximate Nash equilibria for ergodic Mean-Field Games (MFG) in finite state-action spaces. Building on the well-established performance of TRPO in the reinforcement learning (RL) setting, we extend its methodology to the MFG framework, leveraging its stability and robustness in policy optimization. Under standard assumptions in the MFG literature, we provide a rigorous analysis of MF-TRPO, establishing theoretical guarantees on its convergence. Our results cover both the exact formulation of the algorithm and its sample-based counterpart, where we derive high-probability guarantees and finite sample complexity. This work advances MFG optimization by bridging RL techniques with mean-field decision-making, offering a theoretically grounded approach to solving complex multi-agent problems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:50:25 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Ocello",
"Antonio",
""
],
[
"Tiapkin",
"Daniil",
""
],
[
"Mancini",
"Lorenzo",
""
],
[
"Laurière",
"Mathieu",
""
],
[
"Moulines",
"Eric",
""
]
] |
2505.22782 | Mytraya Gattu | Mytraya Gattu and J. K. Jain | Molecular anyons in fractional quantum Hall effect | null | null | null | null | cond-mat.str-el cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | One of the profound consequences of the fractional quantum Hall (FQH) effect is the notion of fractionally charged anyons. In spite of extensive experimental study, puzzles remain, however. For example, both shot-noise and Aharonov-Bohm interference measurements sometimes report a charge that is a multiple of the elementary charge. We report here high-precision microscopic calculations that reveal the surprising result that the FQH anyons often bind together into stable clusters, which we term molecular anyons. This is counterintuitive, given that the elementary anyons carry the same charge and are therefore expected to repel one another. The number of anyons in a cluster, its binding energy and its size depend sensitively on the parent FQH state and the interaction between electrons (which is experimentally tunable, e.g., by varying the quantum well width). Our calculations further suggest that the charge-$1/4$ non-Abelian anyons of the $5/2$ FQH state may also bind to form charge-$1/2$ Abelian clusters. The existence of molecular anyons not only can provide a natural explanation for the observed charges, but also leads to a host of new predictions for future experiments and invites a re-analysis of many past ones. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:51:30 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Gattu",
"Mytraya",
""
],
[
"Jain",
"J. K.",
""
]
] |
2505.22783 | Charles E Thornton | Charles E. Thornton, Jamie Sloop, Samuel Brown, Aaron Orndorff, William C. Headley, Stephen Young | Temporal Convolutional Autoencoder for Interference Mitigation in FMCW Radar Altimeters | 10 pages, 10 figures | null | null | null | eess.SP cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the end-to-end altitude estimation performance of a convolutional autoencoder-based interference mitigation approach for frequency-modulated continuous-wave (FMCW) radar altimeters. Specifically, we show that a Temporal Convolutional Network (TCN) autoencoder effectively exploits temporal correlations in the received signal, providing superior interference suppression compared to a Least Mean Squares (LMS) adaptive filter. Unlike existing approaches, the present method operates directly on the received FMCW signal. Additionally, we identify key challenges in applying deep learning to wideband FMCW interference mitigation and outline directions for future research to enhance real-time feasibility and generalization to arbitrary interference conditions. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:52:10 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Thornton",
"Charles E.",
""
],
[
"Sloop",
"Jamie",
""
],
[
"Brown",
"Samuel",
""
],
[
"Orndorff",
"Aaron",
""
],
[
"Headley",
"William C.",
""
],
[
"Young",
"Stephen",
""
]
] |
2505.22784 | Viraj Nadkarni | Viraj Nadkarni, Pramod Viswanath | Split the Yield, Share the Risk: Pricing, Hedging and Fixed rates in DeFi | null | null | null | null | econ.TH cs.SY eess.SY | http://creativecommons.org/licenses/by/4.0/ | We present the first formal treatment of \emph{yield tokenization}, a mechanism that decomposes yield-bearing assets into principal and yield components to facilitate risk transfer and price discovery in decentralized finance (DeFi). We propose a model that characterizes yield token dynamics using stochastic differential equations. We derive a no-arbitrage pricing framework for yield tokens, enabling their use in hedging future yield volatility and managing interest rate risk in decentralized lending pools. Taking DeFi lending as our focus, we show how both borrowers and lenders can use yield tokens to achieve optimal hedging outcomes and mitigate exposure to adversarial interest rate manipulation. Furthermore, we design automated market makers (AMMs) that incorporate a menu of bonding curves to aggregate liquidity from participants with heterogeneous risk preferences. This leads to an efficient and incentive-compatible mechanism for trading yield tokens and yield futures. Building on these foundations, we propose a modular \textit{fixed-rate} lending protocol that synthesizes on-chain yield token markets and lending pools, enabling robust interest rate discovery and enhancing capital efficiency. Our work provides the theoretical underpinnings for risk management and fixed-income infrastructure in DeFi, offering practical mechanisms for stable and sustainable yield markets. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:57:28 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Nadkarni",
"Viraj",
""
],
[
"Viswanath",
"Pramod",
""
]
] |
2505.22785 | Marco Fumero | Marco Fumero, Luca Moschella, Emanuele Rodol\`a, Francesco Locatello | Navigating the Latent Space Dynamics of Neural Models | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Neural networks transform high-dimensional data into compact, structured representations, often modeled as elements of a lower dimensional latent space. In this paper, we present an alternative interpretation of neural models as dynamical systems acting on the latent manifold. Specifically, we show that autoencoder models implicitly define a latent vector field on the manifold, derived by iteratively applying the encoding-decoding map, without any additional training. We observe that standard training procedures introduce inductive biases that lead to the emergence of attractor points within this vector field. Drawing on this insight, we propose to leverage the vector field as a representation for the network, providing a novel tool to analyze the properties of the model and the data. This representation enables to: (i) analyze the generalization and memorization regimes of neural models, even throughout training; (ii) extract prior knowledge encoded in the network's parameters from the attractors, without requiring any input data; (iii) identify out-of-distribution samples from their trajectories in the vector field. We further validate our approach on vision foundation models, showcasing the applicability and effectiveness of our method in real-world scenarios. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:57:41 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Fumero",
"Marco",
""
],
[
"Moschella",
"Luca",
""
],
[
"Rodolà",
"Emanuele",
""
],
[
"Locatello",
"Francesco",
""
]
] |
2505.22786 | Xiang Liu | Xiang Liu, Junjie Wee and Guo-Wei Wei | Topological Machine Learning for Protein-Nucleic Acid Binding Affinity Changes Upon Mutation | null | null | null | null | q-bio.QM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Understanding how protein mutations affect protein-nucleic acid binding is critical for unraveling disease mechanisms and advancing therapies. Current experimental approaches are laborious, and computational methods remain limited in accuracy. To address this challenge, we propose a novel topological machine learning model (TopoML) combining persistent Laplacian (from topological data analysis) with multi-perspective features: physicochemical properties, topological structures, and protein Transformer-derived sequence embeddings. This integrative framework captures robust representations of protein-nucleic acid binding interactions. To validate the proposed method, we employ two datasets, a protein-DNA dataset with 596 single-point amino acid mutations, and a protein-RNA dataset with 710 single-point amino acid mutations. We show that the proposed TopoML model outperforms state-of-the-art methods in predicting mutation-induced binding affinity changes for protein-DNA and protein-RNA complexes. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:57:51 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liu",
"Xiang",
""
],
[
"Wee",
"Junjie",
""
],
[
"Wei",
"Guo-Wei",
""
]
] |
2505.22787 | Alejandro Lozano | Christopher Polzak, Alejandro Lozano, Min Woo Sun, James Burgess, Yuhui Zhang, Kevin Wu, and Serena Yeung-Levy | Can Large Language Models Match the Conclusions of Systematic Reviews? | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Systematic reviews (SR), in which experts summarize and analyze evidence across individual studies to provide insights on a specialized topic, are a cornerstone for evidence-based clinical decision-making, research, and policy. Given the exponential growth of scientific articles, there is growing interest in using large language models (LLMs) to automate SR generation. However, the ability of LLMs to critically assess evidence and reason across multiple documents to provide recommendations at the same proficiency as domain experts remains poorly characterized. We therefore ask: Can LLMs match the conclusions of systematic reviews written by clinical experts when given access to the same studies? To explore this question, we present MedEvidence, a benchmark pairing findings from 100 SRs with the studies they are based on. We benchmark 24 LLMs on MedEvidence, including reasoning, non-reasoning, medical specialist, and models across varying sizes (from 7B-700B). Through our systematic evaluation, we find that reasoning does not necessarily improve performance, larger models do not consistently yield greater gains, and knowledge-based fine-tuning degrades accuracy on MedEvidence. Instead, most models exhibit similar behavior: performance tends to degrade as token length increases, their responses show overconfidence, and, contrary to human experts, all models show a lack of scientific skepticism toward low-quality findings. These results suggest that more work is still required before LLMs can reliably match the observations from expert-conducted SRs, even though these systems are already deployed and being used by clinicians. We release our codebase and benchmark to the broader research community to further investigate LLM-based SR systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:58:09 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Polzak",
"Christopher",
""
],
[
"Lozano",
"Alejandro",
""
],
[
"Sun",
"Min Woo",
""
],
[
"Burgess",
"James",
""
],
[
"Zhang",
"Yuhui",
""
],
[
"Wu",
"Kevin",
""
],
[
"Yeung-Levy",
"Serena",
""
]
] |
2505.22788 | Nathaniel Wei | Nathaniel J. Wei, Adina Y. Fleisher, John W. Kurelek, Marcus N. Hultmark | Effects of thrust, tip-speed ratio, and time variations on wind-turbine wakes at high Reynolds numbers | null | null | null | null | physics.flu-dyn | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The evolution of rotor wakes is an important problem for a wide range of wind-energy and aerodynamic applications, and is of particular relevance to dynamic wake control strategies for wind farms. This study aims to clarify the influence of turbine thrust and tip-speed ratio on tip-vortex breakdown and the evolution of the near wake. Scaling arguments show that these parameters contribute to the wake dynamics in distinct ways, and that neither thrust nor tip-speed ratio are alone sufficient to describe near and intermediate wake development. These considerations are especially critical for time-varying flows. To demonstrate these principles, a wind turbine at a near utility-scale Reynolds number ($Re_D=4\times10^6$) is forced in periodic rotation-rate oscillations at low Strouhal numbers ($St=0.04$). The slow time-varying forcing protocol decouples thrust and tip-speed ratio effects without introducing nonlinear dynamics into the wake that would appear at higher forcing frequencies. Flow measurements in the wake of the turbine show disturbances propagate through the wake as traveling waves, with thrust and tip-speed ratio variations displaying synergistic or competing effects on wake dynamics depending on the relative phase and amplitude of such disturbances. The results provide key insights into the dynamics of time-varying wakes, limitations in existing models of rotor wake dynamics, and to future novel wake-control schemes. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:01:13 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wei",
"Nathaniel J.",
""
],
[
"Fleisher",
"Adina Y.",
""
],
[
"Kurelek",
"John W.",
""
],
[
"Hultmark",
"Marcus N.",
""
]
] |
2505.22789 | Erbing Hua | Erbing Hua, Theofilos Spyrou, Majid Ahmadi, Abdul Momin Syed, Hanzhi Xun, Laurentiu Braic, Ewout van der Veer, Nazek Elatab, Anteneh Gebregiorgis, Georgi Gaydadjiev, Beatriz Noheda, Said Hamdioui, Ryoichi Ishihara, Heba Abunahla | PdNeuRAM: Forming-Free, Multi-Bit Pd/HfO2 ReRAM for Energy-Efficient Computing | 32 pages, 6 figures in main text and 7 figures in supporting information | null | null | null | cond-mat.mtrl-sci eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Memristor technology shows great promise for energy-efficient computing, yet it grapples with challenges like resistance drift and inherent variability. For filamentary Resistive RAM (ReRAM), one of the most investigated types of memristive devices, the expensive electroforming step required to create conductive pathways results in increased power and area overheads and reduced endurance. In this study, we present novel HfO2-based forming-free ReRAM devices, PdNeuRAM, that operate at low voltages, support multi-bit functionality, and display reduced variability. Through a deep understanding and comprehensive material characterization, we discover the key process that allows this unique behavior: a Pd-O-Hf configuration that capitalizes on Pd innate affinity for integrating into HfO2. This structure actively facilitates charge redistribution at room temperature, effectively eliminating the need for electroforming. Moreover, the fabricated ReRAM device provides tunable resistance states for dense memory and reduces programming and reading energy by 43% and 73%, respectively, using spiking neural networks (SNN). This study reveals novel mechanistic insights and delineates a strategic roadmap for the realization of power-efficient and cost-effective ReRAM devices. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:02:24 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hua",
"Erbing",
""
],
[
"Spyrou",
"Theofilos",
""
],
[
"Ahmadi",
"Majid",
""
],
[
"Syed",
"Abdul Momin",
""
],
[
"Xun",
"Hanzhi",
""
],
[
"Braic",
"Laurentiu",
""
],
[
"van der Veer",
"Ewout",
""
],
[
"Elatab",
"Nazek",
""
],
[
"Gebregiorgis",
"Anteneh",
""
],
[
"Gaydadjiev",
"Georgi",
""
],
[
"Noheda",
"Beatriz",
""
],
[
"Hamdioui",
"Said",
""
],
[
"Ishihara",
"Ryoichi",
""
],
[
"Abunahla",
"Heba",
""
]
] |
2505.22790 | Istv\'an Szanyi | Istv\'an Szanyi | Odderon Exchange in Elastic Proton-Proton and Proton-Antiproton Scattering at TeV Energies | 218 pages, 78 figures; PhD dissertation (thesis) defended at ELTE E\"otv\"os Lor\'and University on 27 March 2025 | null | 10.15476/ELTE.2024.155 | null | hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The odderon, a leading crossing-odd $t$-channel exchange, was first proposed by L. Lukaszuk and B. Nicolescu in 1973, but its existence remained elusive for 48 years. Elastic proton-proton scattering measurements at CERN's Large Hadron Collider (LHC) and elastic proton-antiproton scattering measurements at FNAL's Tevatron, performed at TeV-scale center-of-mass energies ($\sqrt{s}$) and over wide ranges of squared four-momentum transfer ($t$), opened new opportunities to study the physics of elastic hadronic processes which ultimately led to the discovery of odderon exchange in the nonperturbative domain of strong interactions, as detailed in this dissertation. In quantum chromodynamics (QCD), the fundamental theory of strong interactions, the odderon is described in the perturbative regime as a $t$-channel exchange of an odd number of interacting gluons. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:02:40 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Szanyi",
"István",
""
]
] |
2505.22791 | Francesco Libbi | Francesco Libbi, Lorenzo Monacelli, and Boris Kozinsky | Quantum cooling below absolute zero | null | null | null | null | quant-ph cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Similar to other perovskites in its family, SrTiO$_3$ exhibits a significant softening of the ferroelectric mode with decreasing temperature, a behavior that typically heralds the onset of a ferroelectric transition. However, this material remains paraelectric down to 0K due to the Heisenberg quantum uncertainty principle that prevents atoms from localizing in the ferroelectric minima, a fundamental limit that appears impossible to overcome. This work shows that in the strong out-of-equilibrium regime induced by resonant mid-IR pulses, quantum fluctuations can be suppressed, effectively cooling the material below the 0K ground state and inducing a ferroelectric transition in SrTiO$_3$ that is otherwise impossible. The appearance of a metastable state, that is distinct from the conventional ground state, is the first demonstration of how it is possible to leverage and control quantum fluctuations with pulsed light to qualitatively alter the free energy landscape of a quantum system. We predict the conditions and system parameter under which induced non-equilibrium state can be long-lived and even permanent. In providing a quantitative description, based on first principles machine learned potential energy surface, we explain recent experimental observations of light-induced ferroelectric transition in this material. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:03:30 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Libbi",
"Francesco",
""
],
[
"Monacelli",
"Lorenzo",
""
],
[
"Kozinsky",
"Boris",
""
]
] |
2505.22792 | Yuxi Zhang | Yuxi Zhang, Yueting Li, Xinyu Du, Sibo Wang | Rhetorical Text-to-Image Generation via Two-layer Diffusion Policy Optimization | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Generating images from rhetorical languages remains a critical challenge for text-to-image models. Even state-of-the-art (SOTA) multimodal large language models (MLLM) fail to generate images based on the hidden meaning inherent in rhetorical language--despite such content being readily mappable to visual representations by humans. A key limitation is that current models emphasize object-level word embedding alignment, causing metaphorical expressions to steer image generation towards their literal visuals and overlook the intended semantic meaning. To address this, we propose Rhet2Pix, a framework that formulates rhetorical text-to-image generation as a multi-step policy optimization problem, incorporating a two-layer MDP diffusion module. In the outer layer, Rhet2Pix converts the input prompt into incrementally elaborated sub-sentences and executes corresponding image-generation actions, constructing semantically richer visuals. In the inner layer, Rhet2Pix mitigates reward sparsity during image generation by discounting the final reward and optimizing every adjacent action pair along the diffusion denoising trajectory. Extensive experiments demonstrate the effectiveness of Rhet2Pix in rhetorical text-to-image generation. Our model outperforms SOTA MLLMs such as GPT-4o, Grok-3 and leading academic baselines across both qualitative and quantitative evaluations. The code and dataset used in this work are publicly available. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:03:37 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Yuxi",
""
],
[
"Li",
"Yueting",
""
],
[
"Du",
"Xinyu",
""
],
[
"Wang",
"Sibo",
""
]
] |
2505.22793 | Srishti Yadav | Srishti Yadav, Lauren Tilton, Maria Antoniak, Taylor Arnold, Jiaang Li, Siddhesh Milind Pawar, Antonia Karamolegkou, Stella Frank, Zhaochong An, Negar Rostamzadeh, Daniel Hershcovich, Serge Belongie, Ekaterina Shutova | Cultural Evaluations of Vision-Language Models Have a Lot to Learn from Cultural Theory | null | null | null | null | cs.CV cs.CL | http://creativecommons.org/licenses/by/4.0/ | Modern vision-language models (VLMs) often fail at cultural competency evaluations and benchmarks. Given the diversity of applications built upon VLMs, there is renewed interest in understanding how they encode cultural nuances. While individual aspects of this problem have been studied, we still lack a comprehensive framework for systematically identifying and annotating the nuanced cultural dimensions present in images for VLMs. This position paper argues that foundational methodologies from visual culture studies (cultural studies, semiotics, and visual studies) are necessary for cultural analysis of images. Building upon this review, we propose a set of five frameworks, corresponding to cultural dimensions, that must be considered for a more complete analysis of the cultural competencies of VLMs. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:04:04 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yadav",
"Srishti",
""
],
[
"Tilton",
"Lauren",
""
],
[
"Antoniak",
"Maria",
""
],
[
"Arnold",
"Taylor",
""
],
[
"Li",
"Jiaang",
""
],
[
"Pawar",
"Siddhesh Milind",
""
],
[
"Karamolegkou",
"Antonia",
""
],
[
"Frank",
"Stella",
""
],
[
"An",
"Zhaochong",
""
],
[
"Rostamzadeh",
"Negar",
""
],
[
"Hershcovich",
"Daniel",
""
],
[
"Belongie",
"Serge",
""
],
[
"Shutova",
"Ekaterina",
""
]
] |
2505.22794 | Fidele Twagirayezu | Fidele J. Twagirayezu | Timelike Quantum Energy Teleportation in the Nambu-Jona-Lasinio Model | 10 pages, 1 figure(s), 3 tables | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | We propose a novel timelike quantum energy teleportation (QET) protocol within the 1+1 dimensional Nambu-Jona-Lasinio (NJL) model, an interacting fermionic field theory exhibiting spontaneous chiral symmetry breaking. By coupling localized Unruh-DeWitt detectors to the fermionic field, we demonstrate how an initial observer's measurement enables a second observer to extract energy at a later time using only classical information transfer. This protocol leverages the NJL vacuum's rich entanglement structure, driven by the chiral condensate, to facilitate energy transfer without physical particle transport. We derive the energy flows and explore the roles of measurement and time evolution and validate the protocol through quantum circuit simulations on a lattice-regularized NJL model. Our findings highlight the NJL model's potential as a platform for exploring QET in interacting quantum field theories and pave the way for experimental realizations on quantum hardware. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:05:44 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Twagirayezu",
"Fidele J.",
""
]
] |
2505.22795 | Christopher Monahan | Joe Karpie, Christopher J. Monahan, Kostas Orginos, Savvas Zafeiropoulos | Window observables for benchmarking parton distribution functions | 10 pages, 9 figures | null | null | JLAB-THY-25-4347 | hep-lat hep-ph | http://creativecommons.org/licenses/by/4.0/ | Global analysis of collider and fixed-target experimental data and calculations from lattice quantum chromodynamics (QCD) are used to gain complementary information on the structure of hadrons. We propose novel ``window observables'' that allow for higher precision cross-validation between the different approaches, a critical step for studies that wish to combine the datasets. Global analyses are limited by the kinematic regions accessible to experiment, particularly in a range of Bjorken-$x$, and lattice QCD calculations also have limitations requiring extrapolations to obtain the parton distributions. We provide two different ``window observables'' that can be defined within a region of $x$ where extrapolations and interpolations in global analyses remain reliable and where lattice QCD results retain sensitivity and precision. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:11:27 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Karpie",
"Joe",
""
],
[
"Monahan",
"Christopher J.",
""
],
[
"Orginos",
"Kostas",
""
],
[
"Zafeiropoulos",
"Savvas",
""
]
] |
2505.22796 | Benjamin Ueland | S. X. M. Riberolles, A. M. Nedi\'c, B. Kuthanazhi, F. Ye, S. L. Bud'ko, P. C. Canfield, R. J. McQueeney, Junyeong Ahn, V. L. Quito, T. V. Trevisan, L. L. Wang, P. P. Orth, B. G. Ueland | Symmetry tuning topological states of an axion insulator with noncollinear magnetic order | null | null | null | null | cond-mat.mtrl-sci cond-mat.str-el | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Topological properties of quantum materials are intimately related to symmetry. Here, we tune the magnetic order of the axion insulator candidate EuIn$_2$As$_2$ from its broken-helix ground state to the field-polarized phase by applying an in-plane magnetic field. Using results from neutron diffraction and magnetization measurements with ab inito theory and symmetry analysis, we determine how the field tunes the magnetic symmetry within individual magnetic domains and examine the resulting changes to the topological surface states and hinge states existing on edges shared by certain surfaces hosting gapped Dirac states. We predict field-tunable complex and domain-specific hinge-state patterns, with some crystal surfaces undergoing a field-induced topological phase transition. We further find that domain walls have pinned hinge states when intersecting certain crystal surfaces, providing another channel for tuning the chiral-charge-transport pathways. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:12:18 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Riberolles",
"S. X. M.",
""
],
[
"Nedić",
"A. M.",
""
],
[
"Kuthanazhi",
"B.",
""
],
[
"Ye",
"F.",
""
],
[
"Bud'ko",
"S. L.",
""
],
[
"Canfield",
"P. C.",
""
],
[
"McQueeney",
"R. J.",
""
],
[
"Ahn",
"Junyeong",
""
],
[
"Quito",
"V. L.",
""
],
[
"Trevisan",
"T. V.",
""
],
[
"Wang",
"L. L.",
""
],
[
"Orth",
"P. P.",
""
],
[
"Ueland",
"B. G.",
""
]
] |
2505.22797 | Vladyslav Gapyak | Vladyslav Gapyak, Thomas M\"arz, Andreas Weinmann | Fast Trajectory-Independent Model-Based Reconstruction Algorithm for Multi-Dimensional Magnetic Particle Imaging | 10 pages, 5 figures. This work has been submitted to the IEEE for possible publication | null | null | null | cs.CV cs.NA math.NA physics.med-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Magnetic Particle Imaging (MPI) is a promising tomographic technique for visualizing the spatio-temporal distribution of superparamagnetic nanoparticles, with applications ranging from cancer detection to real-time cardiovascular monitoring. Traditional MPI reconstruction relies on either time-consuming calibration (measured system matrix) or model-based simulation of the forward operator. Recent developments have shown the applicability of Chebyshev polynomials to multi-dimensional Lissajous Field-Free Point (FFP) scans. This method is bound to the particular choice of sinusoidal scanning trajectories. In this paper, we present the first reconstruction on real 2D MPI data with a trajectory-independent model-based MPI reconstruction algorithm. We further develop the zero-shot Plug-and-Play (PnP) algorithm of the authors -- with automatic noise level estimation -- to address the present deconvolution problem, leveraging a state-of-the-art denoiser trained on natural images without retraining on MPI-specific data. We evaluate our method on the publicly available 2D FFP MPI dataset ``MPIdata: Equilibrium Model with Anisotropy", featuring scans of six phantoms acquired using a Bruker preclinical scanner. Moreover, we show reconstruction performed on custom data on a 2D scanner with additional high-frequency excitation field and partial data. Our results demonstrate strong reconstruction capabilities across different scanning scenarios -- setting a precedent for general-purpose, flexible model-based MPI reconstruction. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:13:46 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Gapyak",
"Vladyslav",
""
],
[
"März",
"Thomas",
""
],
[
"Weinmann",
"Andreas",
""
]
] |
2505.22798 | Anton Bj\"orklund | Anton Bj\"orklund, Mykola Zaitsev, Marta Kwiatkowska | Efficient Preimage Approximation for Neural Network Certification | null | null | null | null | cs.LG cs.AI cs.CR | http://creativecommons.org/licenses/by-sa/4.0/ | The growing reliance on artificial intelligence in safety- and security-critical applications demands effective neural network certification. A challenging real-world use case is certification against ``patch attacks'', where adversarial patches or lighting conditions obscure parts of images, for example traffic signs. One approach to certification, which also gives quantitative coverage estimates, utilizes preimages of neural networks, i.e., the set of inputs that lead to a specified output. However, these preimage approximation methods, including the state-of-the-art PREMAP algorithm, struggle with scalability. This paper presents novel algorithmic improvements to PREMAP involving tighter bounds, adaptive Monte Carlo sampling, and improved branching heuristics. We demonstrate efficiency improvements of at least an order of magnitude on reinforcement learning control benchmarks, and show that our method scales to convolutional neural networks that were previously infeasible. Our results demonstrate the potential of preimage approximation methodology for reliability and robustness certification. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:13:56 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Björklund",
"Anton",
""
],
[
"Zaitsev",
"Mykola",
""
],
[
"Kwiatkowska",
"Marta",
""
]
] |
2505.22799 | Ivan Christov | Uday M. Rade, Shrihari D. Pande, Ivan C. Christov | Theory and simulation of elastoinertial rectification of oscillatory flows in two-dimensional deformable rectangular channels | 23 pages, 13 figures | null | null | null | physics.flu-dyn | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A slender two-dimensional (2D) channel bounded by a rigid bottom surface and a slender elastic layer above deforms when a fluid flows through it. Hydrodynamic forces cause deformation at the fluid-solid interface, which in turn changes the cross-sectional area of the fluidic channel. The nonlinear coupling between flow and deformation, along with the attendant asymmetry in geometry caused by flow-induced deformation, produces a streaming effect (a non-zero cycle-average despite time-periodic forcing). Surprisingly, fluid inertia provides another nonlinear coupling, tightly connected to deformation, that enhances streaming, termed ``elastoinertial rectification'' by Zhang and Rallabandi [J. Fluid Mech. 996, A16 (2024)]. We adapt the latter theory of how two-way coupled fluid--structure interaction (FSI) produces streaming to a 2D rectangular configuration, specifically taking care to capture the deformations of the nearly incompressible slender elastic layer via the combined foundation model of Chandler and Vella [Proc. R. Soc. A 476, 20200551 (2020)]. We supplement the elastoinertial rectification theory with direct numerical simulations performed using a conforming arbitrary Lagrangian-Eulerian (ALE) FSI formulation with streamline upwind Petrov-Galerkin stabilization, implemented via the open-source computing platform FEniCS. We examine the axial variation of the cycle-averaged pressure as a function of key dimensionless groups of the problem: the Womersley number, the elastoviscous number, and the compliance number. Assuming a small compliance number, we find excellent agreement between theory and simulations for the leading-order pressure and deformation across a range of conditions. At the next order, the cycle-averaged pressures agree well; however, the cycle-averaged deformation is found to exhibit significant axial and vertical displacements, unlike the combined foundation model. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:14:26 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Rade",
"Uday M.",
""
],
[
"Pande",
"Shrihari D.",
""
],
[
"Christov",
"Ivan C.",
""
]
] |
2505.22800 | Abhishek Bajgain | Abhishek Bajgain, Santu Prasad Jana, Alexander Samokhvalov, Thomas Parker, John Derek Demaree, and Ramesh C. Budhani | Selenization of V2O5/WO3 Bilayers for Tuned Optoelectronic Response of WSe2 Films | 7 pages,5 figures | null | null | null | cond-mat.mes-hall cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Scalable and controlled doping of two-dimensional transition metal dichalcogenides is essential for tuning their electronic and optoelectronic properties. In this work, we demonstrate a robust approach for substitution of vanadium in tungsten diselenide (WSe$_2$) via the selenization of pre-deposited V$_2$O$_5$/WO$_3$ thin films. By adjusting the thickness of the vanadium oxide layer, the V concentration in W$_{1-x}$V$_x$Se$_2$ is systematically varied. Electrical measurements on field-effect transistors reveal a substantial enhancement in hole conduction, with drain current increasing by nearly three orders of magnitude compared to undoped WSe$_2$. Temperature-dependent electrical resistivity indicates a clear insulator-to-metal transition with increasing V content, likely due to band structure modifications. Concurrently, the photoconductive gain decreases, suggesting enhanced recombination and charge screening effects. These results establish vanadium doping via selenization of V$_2$O$_5$/WO$_3$ films as a scalable strategy for modulating the transport and photoresponse of WSe$_2$, offering promising implications for wafer-scale optoelectronic device integration. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:15:17 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Bajgain",
"Abhishek",
""
],
[
"Jana",
"Santu Prasad",
""
],
[
"Samokhvalov",
"Alexander",
""
],
[
"Parker",
"Thomas",
""
],
[
"Demaree",
"John Derek",
""
],
[
"Budhani",
"Ramesh C.",
""
]
] |
2505.22801 | Qing Wang | Qing Wang, Yuepei Li, Qiao Qiao, Kang Zhou, Qi Li | Towards a More Generalized Approach in Open Relation Extraction | Accepted by ACL 2025 Main Conference | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Open Relation Extraction (OpenRE) seeks to identify and extract novel relational facts between named entities from unlabeled data without pre-defined relation schemas. Traditional OpenRE methods typically assume that the unlabeled data consists solely of novel relations or is pre-divided into known and novel instances. However, in real-world scenarios, novel relations are arbitrarily distributed. In this paper, we propose a generalized OpenRE setting that considers unlabeled data as a mixture of both known and novel instances. To address this, we propose MixORE, a two-phase framework that integrates relation classification and clustering to jointly learn known and novel relations. Experiments on three benchmark datasets demonstrate that MixORE consistently outperforms competitive baselines in known relation classification and novel relation clustering. Our findings contribute to the advancement of generalized OpenRE research and real-world applications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:21:54 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Qing",
""
],
[
"Li",
"Yuepei",
""
],
[
"Qiao",
"Qiao",
""
],
[
"Zhou",
"Kang",
""
],
[
"Li",
"Qi",
""
]
] |
2505.22802 | Tim Evans | Tim S. Evans | From Signed Networks to Group Graphs | 43 pages including 7 in the appendices | null | null | Imperial/TP/25/TSE/1 | physics.soc-ph cs.DM cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | I show that when there is a symmetry in a process defined on the nodes of a network, this can be captured by a new structure, the ``group graph'', in which group elements label the links of a network. I show that group graphs are a generalisation of signed networks which are an example of a $Z_2$ group graph. I also show that the concept of balance in signed networks can be generalised to group graphs. Finally, I show how the dynamics of processes on a consistent group graph are completely controlled by the topology of the underlying network, not by the symmetry group. This generalises recent results on signed networks (Tian and Lambiotte, 2024a) and complex networks (Tian and Lambiotte, 2024b). | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:23:41 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Evans",
"Tim S.",
""
]
] |
2505.22803 | Pedro Mendes | Pedro Mendes, Paolo Romano, David Garlan | CLUE: Neural Networks Calibration via Learning Uncertainty-Error alignment | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Reliable uncertainty estimation is critical for deploying neural networks (NNs) in real-world applications. While existing calibration techniques often rely on post-hoc adjustments or coarse-grained binning methods, they remain limited in scalability, differentiability, and generalization across domains. In this work, we introduce CLUE (Calibration via Learning Uncertainty-Error Alignment), a novel approach that explicitly aligns predicted uncertainty with observed error during training, grounded in the principle that well-calibrated models should produce uncertainty estimates that match their empirical loss. CLUE adopts a novel loss function that jointly optimizes predictive performance and calibration, using summary statistics of uncertainty and loss as proxies. The proposed method is fully differentiable, domain-agnostic, and compatible with standard training pipelines. Through extensive experiments on vision, regression, and language modeling tasks, including out-of-distribution and domain-shift scenarios, we demonstrate that CLUE achieves superior calibration quality and competitive predictive performance with respect to state-of-the-art approaches without imposing significant computational overhead. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:23:47 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Mendes",
"Pedro",
""
],
[
"Romano",
"Paolo",
""
],
[
"Garlan",
"David",
""
]
] |
2505.22804 | Jonghan Lim | Jonghan Lim, Ilya Kovalenko | Dynamic Task Adaptation for Multi-Robot Manufacturing Systems with Large Language Models | null | null | null | null | cs.RO cs.MA | http://creativecommons.org/licenses/by/4.0/ | Recent manufacturing systems are increasingly adopting multi-robot collaboration to handle complex and dynamic environments. While multi-agent architectures support decentralized coordination among robot agents, they often face challenges in enabling real-time adaptability for unexpected disruptions without predefined rules. Recent advances in large language models offer new opportunities for context-aware decision-making to enable adaptive responses to unexpected changes. This paper presents an initial exploratory implementation of a large language model-enabled control framework for dynamic task reassignment in multi-robot manufacturing systems. A central controller agent leverages the large language model's ability to interpret structured robot configuration data and generate valid reassignments in response to robot failures. Experiments in a real-world setup demonstrate high task success rates in recovering from failures, highlighting the potential of this approach to improve adaptability in multi-robot manufacturing systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:26:07 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lim",
"Jonghan",
""
],
[
"Kovalenko",
"Ilya",
""
]
] |
2505.22805 | Siddharth Ancha | Siddharth Ancha, Sunshine Jiang, Travis Manderson, Laura Brandt, Yilun Du, Philip R. Osteen, Nicholas Roy | Anomalies by Synthesis: Anomaly Detection using Generative Diffusion Models for Off-Road Navigation | Presented at ICRA 2025 | null | null | null | cs.RO cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In order to navigate safely and reliably in off-road and unstructured environments, robots must detect anomalies that are out-of-distribution (OOD) with respect to the training data. We present an analysis-by-synthesis approach for pixel-wise anomaly detection without making any assumptions about the nature of OOD data. Given an input image, we use a generative diffusion model to synthesize an edited image that removes anomalies while keeping the remaining image unchanged. Then, we formulate anomaly detection as analyzing which image segments were modified by the diffusion model. We propose a novel inference approach for guided diffusion by analyzing the ideal guidance gradient and deriving a principled approximation that bootstraps the diffusion model to predict guidance gradients. Our editing technique is purely test-time that can be integrated into existing workflows without the need for retraining or fine-tuning. Finally, we use a combination of vision-language foundation models to compare pixels in a learned feature space and detect semantically meaningful edits, enabling accurate anomaly detection for off-road navigation. Project website: https://siddancha.github.io/anomalies-by-diffusion-synthesis/ | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:26:48 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Ancha",
"Siddharth",
""
],
[
"Jiang",
"Sunshine",
""
],
[
"Manderson",
"Travis",
""
],
[
"Brandt",
"Laura",
""
],
[
"Du",
"Yilun",
""
],
[
"Osteen",
"Philip R.",
""
],
[
"Roy",
"Nicholas",
""
]
] |
2505.22806 | Raluca Rufu | Raluca Rufu and Robin M. Canup | Origin of compact exoplanetary systems during disk infall | null | Nat Commun 16, 4853 (2025) | 10.1038/s41467-025-60017-8 | null | astro-ph.EP physics.space-ph | http://creativecommons.org/licenses/by/4.0/ | Exoplanetary systems that contain multiple planets on short-period orbits appear to be prevalent in the current observed exoplanetary population, yet the processes that give rise to such configurations remain poorly understood. A common prior assumption is that planetary accretion commences after the infall of gas and solids to the circumstellar disk ended. However, observational evidence indicates that accretion may begin earlier. We propose that compact systems are surviving remnants of planet accretion that occurred during the final phases of infall. In regions of the disk experiencing ongoing infall, the planetary mass is set by the balance between accretion of infalling solids and the increasingly rapid inward migration driven by the surrounding gas as the planet grows. This balance selects for similarly-sized planets whose mass is a function of infall and disk conditions. We show that infall-produced planets can survive until the gas disk disperses and migration ends, and that across a broad range of conditions, the mass of surviving systems is regulated to a few 10^{-5} to 10^{-4} times the host star's mass. This provides an explanation for the similar mass ratios of known compact systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:27:30 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Rufu",
"Raluca",
""
],
[
"Canup",
"Robin M.",
""
]
] |
2505.22807 | John Duchi | John C. Duchi | Distribution free M-estimation | 26 pages | null | null | null | math.ST cs.IT cs.LG math.IT stat.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The basic question of delineating those statistical problems that are solvable without making any assumptions on the underlying data distribution has long animated statistics and learning theory. This paper characterizes when a (univariate) convex M-estimation or stochastic optimization problem is solvable in such an assumption-free setting, providing a precise dividing line between solvable and unsolvable problems. The conditions we identify show, perhaps surprisingly, that Lipschitz continuity of the loss being minimized is not necessary for distribution free minimization, and they are also distinct from classical characterizations of learnability in machine learning. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:33:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Duchi",
"John C.",
""
]
] |
2505.22808 | Daniel Apai Dr | Daniel Apai, Rory Barnes, Matthew M. Murphy, Tim Lichtenberg, Noah Tuchow, Regis Ferriere, Kevin Wagner, Antonin Affholder, Renu Malhotra, Baptiste Journaux, Allona Vazan, Ramses Ramirez, Abel Mendez, Stephen R. Kane, Veronica H. Klawender, and NExSS Quantitative Habitability Science Working Group | A Terminology and Quantitative Framework for Assessing the Habitability of Solar System and Extraterrestrial Worlds | In press, Planetary Science Journal | null | null | null | astro-ph.EP | http://creativecommons.org/licenses/by/4.0/ | The search for extraterrestrial life in the Solar System and beyond is a key science driver in astrobiology, planetary science, and astrophysics. A critical step is the identification and characterization of potential habitats, both to guide the search and to interpret its results. However, a well-accepted, self-consistent, flexible, and quantitative terminology and method of assessment of habitability are lacking. Our paper fills this gap based on a three year-long study by the NExSS Quantitative Habitability Science Working Group. We reviewed past studies of habitability, but find that the lack of a universally valid definition of life prohibits a universally applicable definition of habitability. A more nuanced approach is needed. We introduce a quantitative habitability assessment framework (QHF) that enables self-consistent, probabilistic assessment of the compatibility of two models: First, a habitat model, which describes the probability distributions of key conditions in the habitat. Second, a viability model, which describes the probability that a metabolism is viable given a set of environmental conditions. We provide an open-source implementation of this framework and four examples as a proof of concept: (a) Comparison of two exoplanets for observational target prioritization; (b) Interpretation of atmospheric O2 detection in two exoplanets; (c) Subsurface habitability of Mars; and (d) Ocean habitability in Europa. These examples demonstrate that our framework can self-consistently inform astrobiology research over a broad range of questions. The proposed framework is modular so that future work can expand the range and complexity of models available, both for habitats and for metabolisms. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:33:51 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Apai",
"Daniel",
""
],
[
"Barnes",
"Rory",
""
],
[
"Murphy",
"Matthew M.",
""
],
[
"Lichtenberg",
"Tim",
""
],
[
"Tuchow",
"Noah",
""
],
[
"Ferriere",
"Regis",
""
],
[
"Wagner",
"Kevin",
""
],
[
"Affholder",
"Antonin",
""
],
[
"Malhotra",
"Renu",
""
],
[
"Journaux",
"Baptiste",
""
],
[
"Vazan",
"Allona",
""
],
[
"Ramirez",
"Ramses",
""
],
[
"Mendez",
"Abel",
""
],
[
"Kane",
"Stephen R.",
""
],
[
"Klawender",
"Veronica H.",
""
],
[
"Group",
"NExSS Quantitative Habitability Science Working",
""
]
] |
2505.22809 | Andrew Zhu | Andrew Zhu, Evan Osgood, Chris Callison-Burch | First Steps Towards Overhearing LLM Agents: A Case Study With Dungeons & Dragons Gameplay | 8 pages, 5 figures. In submission at EMNLP 2025 | null | null | null | cs.CL cs.AI cs.HC | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Much work has been done on conversational LLM agents which directly assist human users with tasks. We present an alternative paradigm for interacting with LLM agents, which we call "overhearing agents". These overhearing agents do not actively participate in conversation -- instead, they "listen in" on human-to-human conversations and perform background tasks or provide suggestions to assist the user. In this work, we explore the overhearing agents paradigm through the lens of Dungeons & Dragons gameplay. We present an in-depth study using large multimodal audio-language models as overhearing agents to assist a Dungeon Master. We perform a human evaluation to examine the helpfulness of such agents and find that some large audio-language models have the emergent ability to perform overhearing agent tasks using implicit audio cues. Finally, we release Python libraries and our project code to support further research into the overhearing agents paradigm at https://github.com/zhudotexe/overhearing_agents. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:34:36 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhu",
"Andrew",
""
],
[
"Osgood",
"Evan",
""
],
[
"Callison-Burch",
"Chris",
""
]
] |
2505.22810 | Yan Shu | Zhoufaran Yang, Yan Shu, Zhifei Yang, Yan Zhang, Yu Li, Keyang Lu, Gangyan Zeng, Shaohui Liu, Yu Zhou, Nicu Sebe | VidText: Towards Comprehensive Evaluation for Video Text Understanding | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Visual texts embedded in videos carry rich semantic information, which is crucial for both holistic video understanding and fine-grained reasoning about local human actions. However, existing video understanding benchmarks largely overlook textual information, while OCR-specific benchmarks are constrained to static images, limiting their ability to capture the interaction between text and dynamic visual contexts. To address this gap, we propose VidText, a new benchmark designed for comprehensive and in-depth evaluation of video text understanding. VidText offers the following key features: 1) It covers a wide range of real-world scenarios and supports multilingual content, encompassing diverse settings where video text naturally appears. 2) It introduces a hierarchical evaluation framework with video-level, clip-level, and instance-level tasks, enabling assessment of both global summarization and local retrieval capabilities. 3) The benchmark also introduces a set of paired perception reasoning tasks, ranging from visual text perception to cross-modal reasoning between textual and visual information. Extensive experiments on 18 state-of-the-art Large Multimodal Models (LMMs) reveal that current models struggle across most tasks, with significant room for improvement. Further analysis highlights the impact of both model-intrinsic factors, such as input resolution and OCR capability, and external factors, including the use of auxiliary information and Chain-of-Thought reasoning strategies. We hope VidText will fill the current gap in video understanding benchmarks and serve as a foundation for future research on multimodal reasoning with video text in dynamic environments. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:39:35 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yang",
"Zhoufaran",
""
],
[
"Shu",
"Yan",
""
],
[
"Yang",
"Zhifei",
""
],
[
"Zhang",
"Yan",
""
],
[
"Li",
"Yu",
""
],
[
"Lu",
"Keyang",
""
],
[
"Zeng",
"Gangyan",
""
],
[
"Liu",
"Shaohui",
""
],
[
"Zhou",
"Yu",
""
],
[
"Sebe",
"Nicu",
""
]
] |
2505.22811 | Ba-Hien Tran | Ba-Hien Tran, Van Minh Nguyen | Highly Efficient and Effective LLMs with Multi-Boolean Architectures | Under Review | null | null | null | stat.ML cs.LG | http://creativecommons.org/licenses/by/4.0/ | Weight binarization has emerged as a promising strategy to drastically reduce the complexity of large language models (LLMs). It is mainly classified into two approaches: post-training binarization and finetuning with training-aware binarization methods. The first approach, while having low complexity, leads to significant loss of information from the original LLMs, resulting in poor performance. The second approach, on the other hand, relies heavily on full-precision latent weights for gradient approximation of binary weights, which not only remains suboptimal but also introduces substantial complexity. In this paper, we introduce a novel framework that effectively transforms LLMs into multi-kernel Boolean parameters, for the first time, finetunes them directly in the Boolean domain, eliminating the need for expensive latent weights. This significantly reduces complexity during both finetuning and inference. Through extensive and insightful experiments across a wide range of LLMs, we demonstrate that our method outperforms recent ultra low-bit quantization and binarization methods. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:40:34 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Tran",
"Ba-Hien",
""
],
[
"Nguyen",
"Van Minh",
""
]
] |
2505.22812 | Ludvig Modin | Ludvig Modin | Moduli spaces for $\Theta$-strata and non-reductive quotients | 32 pages | null | null | null | math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We give a new proof of the $\hat{U}$-theorem of B\'erczi, Doran, Hawes and Kirwan on the existence of geometric quotients for actions of graded unipotent groups in terms of stacks of filtrations and gradings introduced by Halpern-Leistner. Our proof works over any affine Noetherian base, in particular it simultaneously generalizes the previous results to arbitrary characteristic, actions in families and to general $\Theta$-strata. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:40:48 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Modin",
"Ludvig",
""
]
] |
2505.22813 | Josiah Couch | Josiah Couch, Miao Li, Rima Arnaout, Ramy Arnaout | X-Factor: Quality Is a Dataset-Intrinsic Property | 13 pages, 7 figures | null | null | null | cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | In the universal quest to optimize machine-learning classifiers, three factors -- model architecture, dataset size, and class balance -- have been shown to influence test-time performance but do not fully account for it. Previously, evidence was presented for an additional factor that can be referred to as dataset quality, but it was unclear whether this was actually a joint property of the dataset and the model architecture, or an intrinsic property of the dataset itself. If quality is truly dataset-intrinsic and independent of model architecture, dataset size, and class balance, then the same datasets should perform better (or worse) regardless of these other factors. To test this hypothesis, here we create thousands of datasets, each controlled for size and class balance, and use them to train classifiers with a wide range of architectures, from random forests and support-vector machines to deep networks. We find that classifier performance correlates strongly by subset across architectures ($R^2=0.79$), supporting quality as an intrinsic property of datasets independent of dataset size and class balance and of model architecture. Digging deeper, we find that dataset quality appears to be an emergent property of something more fundamental: the quality of datasets' constituent classes. Thus, quality joins size, class balance, and model architecture as an independent correlate of performance and a separate target for optimizing machine-learning-based classification. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:41:37 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Couch",
"Josiah",
""
],
[
"Li",
"Miao",
""
],
[
"Arnaout",
"Rima",
""
],
[
"Arnaout",
"Ramy",
""
]
] |
2505.22814 | Jonghan Lim | Jonghan Lim, Ilya Kovalenko | A Large Language Model-Enabled Control Architecture for Dynamic Resource Capability Exploration in Multi-Agent Manufacturing Systems | null | null | null | null | cs.MA cs.AI | http://creativecommons.org/licenses/by/4.0/ | Manufacturing environments are becoming more complex and unpredictable due to factors such as demand variations and shorter product lifespans. This complexity requires real-time decision-making and adaptation to disruptions. Traditional control approaches highlight the need for advanced control strategies capable of overcoming unforeseen challenges, as they demonstrate limitations in responsiveness within dynamic industrial settings. Multi-agent systems address these challenges through decentralization of decision-making, enabling systems to respond dynamically to operational changes. However, current multi-agent systems encounter challenges related to real-time adaptation, context-aware decision-making, and the dynamic exploration of resource capabilities. Large language models provide the possibility to overcome these limitations through context-aware decision-making capabilities. This paper introduces a large language model-enabled control architecture for multi-agent manufacturing systems to dynamically explore resource capabilities in response to real-time disruptions. A simulation-based case study demonstrates that the proposed architecture improves system resilience and flexibility. The case study findings show improved throughput and efficient resource utilization compared to existing approaches. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:43:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lim",
"Jonghan",
""
],
[
"Kovalenko",
"Ilya",
""
]
] |
2505.22815 | Zhangyi Hu | Zhangyi Hu and Jiemin Wu and Hua Xu and Mingqian Liao and Ninghui Feng and Bo Gao and Songning Lai and Yutao Yue | IMTS is Worth Time $\times$ Channel Patches: Visual Masked Autoencoders for Irregular Multivariate Time Series Prediction | ICML 2025 | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Irregular Multivariate Time Series (IMTS) forecasting is challenging due to the unaligned nature of multi-channel signals and the prevalence of extensive missing data. Existing methods struggle to capture reliable temporal patterns from such data due to significant missing values. While pre-trained foundation models show potential for addressing these challenges, they are typically designed for Regularly Sampled Time Series (RTS). Motivated by the visual Mask AutoEncoder's (MAE) powerful capability for modeling sparse multi-channel information and its success in RTS forecasting, we propose VIMTS, a framework adapting Visual MAE for IMTS forecasting. To mitigate the effect of missing values, VIMTS first processes IMTS along the timeline into feature patches at equal intervals. These patches are then complemented using learned cross-channel dependencies. Then it leverages visual MAE's capability in handling sparse multichannel data for patch reconstruction, followed by a coarse-to-fine technique to generate precise predictions from focused contexts. In addition, we integrate self-supervised learning for improved IMTS modeling by adapting the visual MAE to IMTS data. Extensive experiments demonstrate VIMTS's superior performance and few-shot capability, advancing the application of visual foundation models in more general time series tasks. Our code is available at https://github.com/WHU-HZY/VIMTS. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:44:03 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hu",
"Zhangyi",
""
],
[
"Wu",
"Jiemin",
""
],
[
"Xu",
"Hua",
""
],
[
"Liao",
"Mingqian",
""
],
[
"Feng",
"Ninghui",
""
],
[
"Gao",
"Bo",
""
],
[
"Lai",
"Songning",
""
],
[
"Yue",
"Yutao",
""
]
] |
2505.22816 | Dominik Hahn | Dominik Hahn, S. A. Parameswaran, Benedikt Placke | Provably Efficient Quantum Thermal State Preparation via Local Driving | 28 pages, 3 Figures | null | null | null | quant-ph cond-mat.other cond-mat.stat-mech math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | Preparing the thermal density matrix $\rho_{\beta}\propto e^{-\beta H}$ corresponding to a given Hamiltonian $H$ is a task of central interest across quantum many-body physics, and is particularly salient when attempting to study it with quantum computers. Although solved {in principle} by recent constructions of efficiently simulable Lindblad master equations -- that provably have $\rho_{\beta}$ as a steady state [C.-F. Chen {\it et al}, arXiv:2311.09207] -- their implementation requires large-scale quantum computational resources and is hence challenging {in practice} on current or even near-term quantum devices. Here, we propose a scheme for approximately preparing quantum thermal states that only requires the [repeated] implementation of three readily available ingredients: (a) analog simulation of $H$; (b) strictly local but time-dependent couplings to ancilla qubits; and (c) reset of the ancillas. We give rigorous performance guarantees independent of detailed physical knowledge of $H$ beyond its locality. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:48:07 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hahn",
"Dominik",
""
],
[
"Parameswaran",
"S. A.",
""
],
[
"Placke",
"Benedikt",
""
]
] |
2505.22817 | Robert Hoye | Woo Hyeon Jeong, Junzhi Ye, Jongbeom Kim, Rui Xu, Xinyu Shen, Chia-Yu Chang, Eilidh L. Quinn, Myoung Hoon Song, Peter Nellist, Henry J. Snaith, Yunwei Zhang, Bo Ram Lee and Robert L. Z. Hoye | Enhanced Stability and Linearly Polarized Emission from CsPbI$_3$ Perovskite Nanoplatelets through A-site Cation Engineering | 19 pages, 5 figures | null | null | null | cond-mat.mtrl-sci physics.app-ph | http://creativecommons.org/licenses/by/4.0/ | The anisotropy of perovskite nanoplatelets (PeNPLs) opens up many opportunities in optoelectronics, including enabling the emission of linearly polarized light. But the limited stability of PeNPLs is a pressing challenge, especially for red-emitting CsPbI$_3$. Herein, we address this limitation by alloying FA into the perovskite cuboctahedral site. Unlike Cs/FA alloying in bulk thin films or nonconfined nanocubes, FA incorporation in nanoplatelets requires meticulous control over the reaction conditions, given that nanoplatelets are obtained in kinetically-driven growth regimes instead of thermodynamically-driven conditions. Through in-situ photoluminescence (PL) measurements, we find that excess FA leads to uncontrolled growth, where phase-impurities and nanoplatelets of multiple thicknesses co-exist. Restricting the FA content to up to 25% Cs substitution enables monodisperse PeNPLs, and increases the PL quantum yield (from 53% to 61%), exciton lifetime (from 18 ns to 27 ns), and stability in ambient air (from ~2 days to >7 days) compared to CsPbI$_3$. This arises due to hydrogen bonding between FA and the oleate and oleylammonium ligands, anchoring them to the surface to improve optoelectronic properties and stability. The reduction in non-radiative recombination, improvement in the nanoplatelet aspect ratio, and higher ligand density lead to FA-containing PeNPLs more effectively forming edge-up superlattices, enhancing the PL degree of linear polarization from 5.1% (CsPbI$_3$) to 9.4% (Cs$_{0.75}$FA$_{0.25}$PbI$_3$). These fundamental insights show how the stability limitations of PeNPLs could be addressed, and these materials grown more precisely to improve their performance as polarized light emitters, critical for utilizing them in next-generation display, bioimaging and communications applications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:52:20 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Jeong",
"Woo Hyeon",
""
],
[
"Ye",
"Junzhi",
""
],
[
"Kim",
"Jongbeom",
""
],
[
"Xu",
"Rui",
""
],
[
"Shen",
"Xinyu",
""
],
[
"Chang",
"Chia-Yu",
""
],
[
"Quinn",
"Eilidh L.",
""
],
[
"Song",
"Myoung Hoon",
""
],
[
"Nellist",
"Peter",
""
],
[
"Snaith",
"Henry J.",
""
],
[
"Zhang",
"Yunwei",
""
],
[
"Lee",
"Bo Ram",
""
],
[
"Hoye",
"Robert L. Z.",
""
]
] |
2505.22818 | Linghan Zhong | Linghan Zhong, Samuel Yuan, Jiyang Zhang, Yu Liu, Pengyu Nie, Junyi Jessy Li, Milos Gligoric | A Tool for Generating Exceptional Behavior Tests With Large Language Models | FSE 2025 Demo (Camera Ready) | null | 10.1145/3696630.3728608 | null | cs.SE cs.AI | http://creativecommons.org/licenses/by/4.0/ | Exceptional behavior tests (EBTs) are crucial in software development for verifying that code correctly handles unwanted events and throws appropriate exceptions. However, prior research has shown that developers often prioritize testing "happy paths", e.g., paths without unwanted events over exceptional scenarios. We present exLong, a framework that automatically generates EBTs to address this gap. exLong leverages a large language model (LLM) fine-tuned from CodeLlama and incorporates reasoning about exception-throwing traces, conditional expressions that guard throw statements, and non-exceptional behavior tests that execute similar traces. Our demonstration video illustrates how exLong can effectively assist developers in creating comprehensive EBTs for their project (available at https://youtu.be/Jro8kMgplZk). | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:53:20 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhong",
"Linghan",
""
],
[
"Yuan",
"Samuel",
""
],
[
"Zhang",
"Jiyang",
""
],
[
"Liu",
"Yu",
""
],
[
"Nie",
"Pengyu",
""
],
[
"Li",
"Junyi Jessy",
""
],
[
"Gligoric",
"Milos",
""
]
] |
2505.22819 | Abdelhay Benmoussa | Abdelhay Benmoussa | An Operator Theoretic Derivation of a Bernoulli Stirling Identity: A Novel Use of the Euler Operator and Integration | 4 pages | null | null | null | math.GM | http://creativecommons.org/licenses/by/4.0/ | We present a novel proof of the classical identity relating Bernoulli numbers \( B_n \) and Stirling numbers of the second kind \( S(n,k) \), given by \[ B_n = \sum_{k=1}^n S(n,k)\frac{(-1)^k k!}{k+1}. \] Unlike traditional derivations based on generating functions or purely combinatorial arguments, our method leverages the Euler operator \( \vartheta = t \frac{d}{dt} \) and its algebraic action on analytic functions. By applying this operator repeatedly to a suitably chosen function and integrating both sides, we derive the identity in a way that reveals a deeper operator-theoretic structure underlying the formula. This approach bridges discrete combinatorics and differential operators, offering a fresh perspective on a classical result. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:55:05 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Benmoussa",
"Abdelhay",
""
]
] |
2505.22820 | Ayush Sawarni | Ayush Sawarni, Sahasrajit Sarmasarkar, Vasilis Syrgkanis | Preference Learning with Response Time | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | This paper investigates the integration of response time data into human preference learning frameworks for more effective reward model elicitation. While binary preference data has become fundamental in fine-tuning foundation models, generative AI systems, and other large-scale models, the valuable temporal information inherent in user decision-making remains largely unexploited. We propose novel methodologies to incorporate response time information alongside binary choice data, leveraging the Evidence Accumulation Drift Diffusion (EZ) model, under which response time is informative of the preference strength. We develop Neyman-orthogonal loss functions that achieve oracle convergence rates for reward model learning, matching the theoretical optimal rates that would be attained if the expected response times for each query were known a priori. Our theoretical analysis demonstrates that for linear reward functions, conventional preference learning suffers from error rates that scale exponentially with reward magnitude. In contrast, our response time-augmented approach reduces this to polynomial scaling, representing a significant improvement in sample efficiency. We extend these guarantees to non-parametric reward function spaces, establishing convergence properties for more complex, realistic reward models. Our extensive experiments validate our theoretical findings in the context of preference learning over images. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:55:54 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Sawarni",
"Ayush",
""
],
[
"Sarmasarkar",
"Sahasrajit",
""
],
[
"Syrgkanis",
"Vasilis",
""
]
] |
2505.22821 | Achim Blumensath | Achim Blumensath | Simple Classes of Automatic Structures | null | null | null | null | cs.LO math.LO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study two subclasses of the class of automatic structures: automatic structures of polynomial growth and Presburger structures. We present algebraic characterisations of the groups and the equivalence structures in these two classes. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 19:59:53 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Blumensath",
"Achim",
""
]
] |
2505.22822 | Beomki Yeo | Paul Gessinger, Heather M. Gray, Attila Krasznahorkay, Charles Leggett, Joana Niermann, Andreas Salzburger, Stephen Nicholas Swatman, and Beomki Yeo | traccc: GPU track reconstruction library for HEP experiments | 8 pages, 3 figures | null | null | null | hep-ex | http://creativecommons.org/licenses/by/4.0/ | We present the current development status and progress of traccc, a GPU track reconstruction library developed in the context of the A Common Tracking Software (ACTS) project. traccc implements tracking algorithms used in high energy physics (HEP) experiments, including the Kalman filter based track finding and fitting. We benchmark the software with data simulated by Geant4 to measure the physics and computing performance. We show that the physics performance for GPU and CPU are very close. We also show that the GPUs can achieve higher computational performance than the CPU for sufficiently large events. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:00:51 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Gessinger",
"Paul",
""
],
[
"Gray",
"Heather M.",
""
],
[
"Krasznahorkay",
"Attila",
""
],
[
"Leggett",
"Charles",
""
],
[
"Niermann",
"Joana",
""
],
[
"Salzburger",
"Andreas",
""
],
[
"Swatman",
"Stephen Nicholas",
""
],
[
"Yeo",
"Beomki",
""
]
] |
2505.22823 | Yingming Wang | Yingming Wang, Pepa Atanasova | Self-Critique and Refinement for Faithful Natural Language Explanations | 21 pages, 10 figures, 14 tables | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | With the rapid development of large language models (LLMs), natural language explanations (NLEs) have become increasingly important for understanding model predictions. However, these explanations often fail to faithfully represent the model's actual reasoning process. While existing work has demonstrated that LLMs can self-critique and refine their initial outputs for various tasks, this capability remains unexplored for improving explanation faithfulness. To address this gap, we introduce Self-critique and Refinement for Natural Language Explanations (SR-NLE), a framework that enables models to improve the faithfulness of their own explanations -- specifically, post-hoc NLEs -- through an iterative critique and refinement process without external supervision. Our framework leverages different feedback mechanisms to guide the refinement process, including natural language self-feedback and, notably, a novel feedback approach based on feature attribution that highlights important input words. Our experiments across three datasets and four state-of-the-art LLMs demonstrate that SR-NLE significantly reduces unfaithfulness rates, with our best method achieving an average unfaithfulness rate of 36.02%, compared to 54.81% for baseline -- an absolute reduction of 18.79%. These findings reveal that the investigated LLMs can indeed refine their explanations to better reflect their actual reasoning process, requiring only appropriate guidance through feedback without additional training or fine-tuning. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:08:42 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Yingming",
""
],
[
"Atanasova",
"Pepa",
""
]
] |
2505.22824 | Dongzhe Zheng | Dongzhe Zheng | Constrained Hamiltonian Systems on Observation-Induced Fiber Bundles: Theory of Symmetry and Integrability | This paper establishes the complete mathematical theory underlying the practical framework presented in "Learning Dynamics under Environmental Constraints via Measurement-Induced Bundle Structures" (ICML 2025 (Forty-Second International Conference on Machine Learning) Spotlight, top 2.5% of ~12,000 submissions) | null | null | null | math.DS math-ph math.DG math.MP | http://creativecommons.org/licenses/by/4.0/ | Classical constrained Hamiltonian theory assumes complete observability of system states, but in reality only partial state information is often available. This paper establishes a complete geometric theoretical framework for handling such incompletely observed systems. By introducing the concept of observation-induced fiber bundles, we naturally extend Dirac constraint theory to the fiber bundle setting, unifying the treatment of state constraints and observation constraints. Main results include: (1) Classification of existence conditions for observation fiber bundles based on characteristic class theory; (2) Complete characterization of Poisson structures on fiber bundles and corresponding symplectic reduction theory; (3) Geometric necessary and sufficient conditions for integrability and Lax pair construction; (4) Extension of Noether's theorem under symmetry group actions. The theoretical framework naturally encompasses a wide range of applications from classical mechanics to modern safety-critical control systems, providing a rigorous mathematical foundation for dynamical analysis under incomplete information. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:09:45 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zheng",
"Dongzhe",
""
]
] |
2505.22825 | Michael Klamkin | Michael Klamkin, Mathieu Tanneau, Pascal Van Hentenryck | PGLearn -- An Open-Source Learning Toolkit for Optimal Power Flow | null | null | null | null | cs.LG cs.AI cs.SY eess.SY math.OC | http://creativecommons.org/licenses/by/4.0/ | Machine Learning (ML) techniques for Optimal Power Flow (OPF) problems have recently garnered significant attention, reflecting a broader trend of leveraging ML to approximate and/or accelerate the resolution of complex optimization problems. These developments are necessitated by the increased volatility and scale in energy production for modern and future grids. However, progress in ML for OPF is hindered by the lack of standardized datasets and evaluation metrics, from generating and solving OPF instances, to training and benchmarking machine learning models. To address this challenge, this paper introduces PGLearn, a comprehensive suite of standardized datasets and evaluation tools for ML and OPF. PGLearn provides datasets that are representative of real-life operating conditions, by explicitly capturing both global and local variability in the data generation, and by, for the first time, including time series data for several large-scale systems. In addition, it supports multiple OPF formulations, including AC, DC, and second-order cone formulations. Standardized datasets are made publicly available to democratize access to this field, reduce the burden of data generation, and enable the fair comparison of various methodologies. PGLearn also includes a robust toolkit for training, evaluating, and benchmarking machine learning models for OPF, with the goal of standardizing performance evaluation across the field. By promoting open, standardized datasets and evaluation metrics, PGLearn aims at democratizing and accelerating research and innovation in machine learning applications for optimal power flow problems. Datasets are available for download at https://www.huggingface.co/PGLearn. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:10:04 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Klamkin",
"Michael",
""
],
[
"Tanneau",
"Mathieu",
""
],
[
"Van Hentenryck",
"Pascal",
""
]
] |
2505.22826 | Peter F. Stadler | Christoph Flamm, Daniel Merkle, Peter F. Stadler | Assembly in Directed Hypergraphs | null | null | null | null | cs.DM | http://creativecommons.org/licenses/by/4.0/ | Assembly theory has received considerable attention in the recent past. Here we analyze the formal framework of this model and show that assembly pathways coincide with certain minimal hyperpaths in B-hypergraphs. This makes it possible to generalize the notion of assembly to general chemical reaction systems and to make explicit the connection to rule based models of chemistry, in particular DPO graph rewriting. We observe, furthermore, that assembly theory is closely related to retrosynthetic analysis in chemistry. The assembly index fits seamlessly into a large family of cost measures for directed hyperpath problems that also encompasses cost functions used in computational synthesis planning. This allows to devise a generic approach to compute complexity measures derived from minimal hyperpaths in rule-derived directed hypergraphs using integer linear programming. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:10:22 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Flamm",
"Christoph",
""
],
[
"Merkle",
"Daniel",
""
],
[
"Stadler",
"Peter F.",
""
]
] |
2505.22827 | Moussa Labbadi | Chaimae El Mortajinea, Moussa Labbadib, Adnane Saoudc and Mostafa Bouzia | Learning-Based Robust Fixed-Time Terminal Sliding Mode Control | null | null | null | null | eess.SY cs.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we develop and analyze an integral fixed-time sliding mode control method for a scenario in which the system model is only partially known, utilizing Gaussian processes. We present two theorems on fixed-time convergence. The first theorem addresses the fully known system model, while the second considers situations where the system's drift is approximated utilizing Gaussian processes (GP) for approximating unknown dynamics. Both theorems establish the global fixed-time stability of the closed-loop system. The stability analysis is based on a straightforward quadratic Lyapunov function. Our proposed method outperforms an established adaptive fixed-time sliding mode control approach, especially when ample training data is available. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:10:23 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Mortajinea",
"Chaimae El",
""
],
[
"Labbadib",
"Moussa",
""
],
[
"Saoudc",
"Adnane",
""
],
[
"Bouzia",
"Mostafa",
""
]
] |
2505.22828 | Yisca Kares | Yisca Kares | The range of the des statistic for conjugacy classes in $S_n$ | null | null | null | null | math.CO | http://creativecommons.org/publicdomain/zero/1.0/ | We determine the range of the des statistic on every conjugacy class in the symmetric group $S_n$, prove that the minimum is $1$ (except for the identity class), and show that every intermediate value from $1$ to the maximum value is attained. We also demonstrate a constructive method to achieve every value in the range and discuss its combinatorial implications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:11:03 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kares",
"Yisca",
""
]
] |
2505.22829 | Kenan Tang | Chenruo Liu, Kenan Tang, Yao Qin, Qi Lei | Bridging Distribution Shift and AI Safety: Conceptual and Methodological Synergies | 35 pages | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | This paper bridges distribution shift and AI safety through a comprehensive analysis of their conceptual and methodological synergies. While prior discussions often focus on narrow cases or informal analogies, we establish two types connections between specific causes of distribution shift and fine-grained AI safety issues: (1) methods addressing a specific shift type can help achieve corresponding safety goals, or (2) certain shifts and safety issues can be formally reduced to each other, enabling mutual adaptation of their methods. Our findings provide a unified perspective that encourages fundamental integration between distribution shift and AI safety research. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:11:30 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liu",
"Chenruo",
""
],
[
"Tang",
"Kenan",
""
],
[
"Qin",
"Yao",
""
],
[
"Lei",
"Qi",
""
]
] |
2505.22830 | Alexander Gill | Alexander Gill, Abhilasha Ravichander, Ana Marasovi\'c | What Has Been Lost with Synthetic Evaluation? | 9 pages main, 5 pages reference, 24 pages appendix | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large language models (LLMs) are increasingly used for data generation. However, creating evaluation benchmarks raises the bar for this emerging paradigm. Benchmarks must target specific phenomena, penalize exploiting shortcuts, and be challenging. Through two case studies, we investigate whether LLMs can meet these demands by generating reasoning over-text benchmarks and comparing them to those created through careful crowdsourcing. Specifically, we evaluate both the validity and difficulty of LLM-generated versions of two high-quality reading comprehension datasets: CondaQA, which evaluates reasoning about negation, and DROP, which targets reasoning about quantities. We find that prompting LLMs can produce variants of these datasets that are often valid according to the annotation guidelines, at a fraction of the cost of the original crowdsourcing effort. However, we show that they are less challenging for LLMs than their human-authored counterparts. This finding sheds light on what may have been lost by generating evaluation data with LLMs, and calls for critically reassessing the immediate use of this increasingly prevalent approach to benchmark creation. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:12:32 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Gill",
"Alexander",
""
],
[
"Ravichander",
"Abhilasha",
""
],
[
"Marasović",
"Ana",
""
]
] |
2505.22831 | Peiling Jiang | Peiling Jiang and Haijun Xia | Orca: Browsing at Scale Through User-Driven and AI-Facilitated Orchestration Across Malleable Webpages | null | null | null | null | cs.HC cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Web-based activities are fundamentally distributed across webpages. However, conventional browsers with stacks of tabs fail to support operating and synthesizing large volumes of information across pages. While recent AI systems enable fully automated web browsing and information synthesis, they often diminish user agency and hinder contextual understanding. Therefore, we explore how AI could instead augment users' interactions with content across webpages and mitigate cognitive and manual efforts. Through literature on information tasks and web browsing challenges, and an iterative design process, we present a rich set of novel interactions with our prototype web browser, Orca. Leveraging AI, Orca supports user-driven exploration, operation, organization, and synthesis of web content at scale. To enable browsing at scale, webpages are treated as malleable materials that humans and AI can collaboratively manipulate and compose into a malleable, dynamic, and browser-level workspace. Our evaluation revealed an increased "appetite" for information foraging, enhanced user control, and more flexibility in sensemaking across a broader information landscape on the web. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:13:39 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Jiang",
"Peiling",
""
],
[
"Xia",
"Haijun",
""
]
] |
2505.22832 | Rajan Mehta | Sophia E Marx, Rajan Amit Mehta | 2-Segal sets and pseudomonoids in the bicategory of spans | null | null | null | null | math.CT math.AT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this survey article, we give an introduction to the notion of a 2-Segal set and prove that 2-Segal sets are equivalent to pseudomonoids in the bicategory of spans. The proof utilizes graphical techniques for 2-Segal sets and spans that should be useful in more general settings.
There are procedures for obtaining an associative algebra from a 2-Segal set (satisfying finiteness conditions). We describe these procedures and give several examples of algebras arising from 2-Segal sets.
Wherever possible, we avoid higher category theory so as to make the paper accessible to a wide audience. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:13:44 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Marx",
"Sophia E",
""
],
[
"Mehta",
"Rajan Amit",
""
]
] |
2505.22833 | Chi Hong Yuen | Chi-Hong Yuen | Enhanced Excited State Population and Coherence via Adiabatic Tunneling Ionization and Excitation | null | null | null | null | quant-ph physics.atom-ph physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | Tunneling ionization followed by strong-field excitation leads to important ultrafast phenomena such as charge migration and lasing. Recent theoretical developments suggest that the population of the ionic excited state can be greatly enhanced due to the complex interplay between tunneling and excitation. In this Letter, using an adiabatic approach for both tunneling and excitation, semi-analytical solutions are derived for the population and coherence of a two-level ionic system. This approach removes the strong-field dressing, revealing novel sub-half-cycle processes for excited state population and coherence buildup. It predicts that the excited state population is enhanced by an order of magnitude, independent of the laser wavelength, while coherence amplitude can be boosted by over four orders of magnitude for a multi-cycle pulse. For a single-cycle pulse, it suggests that coherence amplitude decreases rapidly as the wavelength increases. This work introduces a novel framework for generating and controlling the electronic excited state and coherence using intense laser pulses, with applications in strong-field control of chemistry and lasing. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:13:56 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yuen",
"Chi-Hong",
""
]
] |
2505.22834 | Maria Violaris | Maria Violaris | Counterfactuals in Macroscopic Quantum Physics: Irreversibility, Measurement and Locality | PhD thesis | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Can quantum theory be applied on all scales? While there are many arguments for the universality of quantum theory, this question remains a subject of debate. It is unknown how far the existence of macroscopic irreversibility can be derived from or reconciled with time-reversal symmetric quantum dynamics. Furthermore, reasoning about quantum measurements can appear to produce surprising and even paradoxical outcomes. The classical outcomes of quantum measurements are in some contexts deemed to violate the fundamental principle of locality, in particular when considering entanglement and Bell non-locality. Therefore measurement, irreversibility and locality can all appear to challenge the universality of quantum theory. In this thesis we approach these problems using counterfactuals -- statements about the possibility and impossibility of transformations. Using the principles of constructor theory and quantum information theory, we find novel features of quantum thermodynamics relating to irreversibility, information erasure and coherence. We also develop tools to quantify the full implications of non-commutativity of quantum operators in settings where quantum theory is applied universally to measurement devices. This reveals new ways of characterising the quantum information stored in entanglement and quantum branching structure. Our results reinforce the ability of universal quantum theory to consistently describe both microscopic and macroscopic observers and thermodynamic systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:14:17 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Violaris",
"Maria",
""
]
] |
2505.22835 | Sasha Zotine | Sasha Zotine | Computing higher direct images in Macaulay2 | 5 pages | null | null | null | math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This article highlights the ToricHigherDirectImages package in Macaulay2. The central feature is a method for computing (higher) direct images of line bundles under surjective toric morphisms. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:15:22 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zotine",
"Sasha",
""
]
] |
2505.22836 | Gabriel Turinici | Pierre Brugi\`ere and Gabriel Turinici | Model-Free Deep Hedging with Transaction Costs and Light Data Requirements | null | null | null | null | q-fin.MF q-fin.ST | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Option pricing theory, such as the Black and Scholes (1973) model, provides an explicit solution to construct a strategy that perfectly hedges an option in a continuous-time setting. In practice, however, trading occurs in discrete time and often involves transaction costs, making the direct application of continuous-time solutions potentially suboptimal. Previous studies, such as those by Buehler et al. (2018), Buehler et al. (2019) and Cao et al. (2019), have shown that deep learning or reinforcement learning can be used to derive better hedging strategies than those based on continuous-time models. However, these approaches typically rely on a large number of trajectories (of the order of $10^5$ or $10^6$) to train the model. In this work, we show that using as few as 256 trajectories is sufficient to train a neural network that significantly outperforms, in the Geometric Brownian Motion framework, both the classical Black & Scholes formula and the Leland model, which is arguably one of the most effective explicit alternatives for incorporating transaction costs. The ability to train neural networks with such a small number of trajectories suggests the potential for more practical and simple implementation on real-time financial series. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:16:07 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Brugière",
"Pierre",
""
],
[
"Turinici",
"Gabriel",
""
]
] |
2505.22837 | James Brown | Akshat Tandon, James Brown, Kenny Heitritter, Tarini Hardikar, Kanav Setia, Rene Boettcher, Klaus Schertler, Jasper Simon Krauser | Quantum Reservoir Computing for Corrosion Prediction in Aerospace: A Hybrid Approach for Enhanced Material Degradation Forecasting | 6 pages, 5 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | The prediction of material degradation is an important problem to solve in many industries. Environmental conditions, such as humidity and temperature, are important drivers of degradation processes, with corrosion being one of the most prominent ones. Quantum machine learning is a promising research field but suffers from well known deficits such as barren plateaus and measurement overheads. To address this problem, recent research has examined quantum reservoir computing to address time-series prediction tasks. Although a promising idea, developing circuits that are expressive enough while respecting the limited depths available on current devices is challenging. In classical reservoir computing, the onion echo state network model (ESN) [https://doi.org/10.1007/978-3-031-72359-9_9] was introduced to increase the interpretability of the representation structure of the embeddings. This onion ESN model utilizes a concatenation of smaller reservoirs that describe different time scales by covering different regions of the eigenvalue spectrum. Here, we use the same idea in the realm of quantum reservoir computing by simultaneously evolving smaller quantum reservoirs to better capture all the relevant time-scales while keeping the circuit depth small. We do this by modifying the rotation angles which we show alters the eigenvalues of the quantum evolution, but also note that modifying the number of mid-circuit measurements accomplishes the same goals of changing the long-term or short-term memory. This onion QRC outperforms a simple model and a single classical reservoir for predicting the degradation of aluminum alloys in different environmental conditions. By combining the onion QRC with an additional classical reservoir layer, the prediction accuracy is further improved. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:16:20 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Tandon",
"Akshat",
""
],
[
"Brown",
"James",
""
],
[
"Heitritter",
"Kenny",
""
],
[
"Hardikar",
"Tarini",
""
],
[
"Setia",
"Kanav",
""
],
[
"Boettcher",
"Rene",
""
],
[
"Schertler",
"Klaus",
""
],
[
"Krauser",
"Jasper Simon",
""
]
] |
2505.22838 | Douglas Stinson | Douglas R. Stinson | A Unified Treatment of Some Classic Combinatorial Inequalities Using the Variance Method | null | null | null | null | math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The "variance method" has been used to prove many classical inequalities in design theory and coding theory. The purpose of this expository note is to review and present some of these inequalities in a unified setting. I will also discuss some examples from my own research where I have employed these techniques. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:17:09 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Stinson",
"Douglas R.",
""
]
] |
2505.22839 | Liu Yuezhang | Liu Yuezhang and Xue-Xin Wei | How Do Diffusion Models Improve Adversarial Robustness? | 18 pages, 7 figures | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent findings suggest that diffusion models significantly enhance empirical adversarial robustness. While some intuitive explanations have been proposed, the precise mechanisms underlying these improvements remain unclear. In this work, we systematically investigate how and how well diffusion models improve adversarial robustness. First, we observe that diffusion models intriguingly increase, rather than decrease, the $\ell_p$ distance to clean samples--challenging the intuition that purification denoises inputs closer to the original data. Second, we find that the purified images are heavily influenced by the internal randomness of diffusion models, where a compression effect arises within each randomness configuration. Motivated by this observation, we evaluate robustness under fixed randomness and find that the improvement drops to approximately 24% on CIFAR-10--substantially lower than prior reports approaching 70%. Importantly, we show that this remaining robustness gain strongly correlates with the model's ability to compress the input space, revealing the compression rate as a reliable robustness indicator without requiring gradient-based analysis. Our findings provide novel insights into the mechanisms underlying diffusion-based purification, and offer guidance for developing more effective and principled adversarial purification systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:19:21 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yuezhang",
"Liu",
""
],
[
"Wei",
"Xue-Xin",
""
]
] |
2505.22840 | Mahesh Banavar | Dharambir Mahto, Prashant Yadav, Mahesh Banavar, Jim Keany, Alan T Joseph, Srinivas Kilambi | Development and Validation of SXI++ LNM Algorithm for Sepsis Prediction | Paper accepted at JMAI | D. Mahto, P. Yadav, et. al., "Advancements in early sepsis detection: Predictive performance of SXI++ LNM and COMPOSER algorithms for early sepsis outcome," J. Med. Artif. Intell., vol. 8, p. 22, 2025 | 10.21037/jmai-24-393 | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Sepsis is a life-threatening condition affecting over 48.9 million people globally and causing 11 million deaths annually. Despite medical advancements, predicting sepsis remains a challenge due to non-specific symptoms and complex pathophysiology. The SXI++ LNM is a machine learning scoring system that refines sepsis prediction by leveraging multiple algorithms and deep neural networks. This study aims to improve robustness in clinical applications and evaluates the predictive performance of the SXI++ LNM for sepsis prediction. The model, utilizing a deep neural network, was trained and tested using multiple scenarios with different dataset distributions. The model's performance was assessed against unseen test data, and accuracy, precision, and area under the curve (AUC) were calculated. THE SXI++ LNM outperformed the state of the art in three use cases, achieving an AUC of 0.99 (95% CI: 0.98-1.00). The model demonstrated a precision of 99.9% (95% CI: 99.8-100.0) and an accuracy of 99.99% (95% CI: 99.98-100.0), maintaining high reliability. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:20:35 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Mahto",
"Dharambir",
""
],
[
"Yadav",
"Prashant",
""
],
[
"Banavar",
"Mahesh",
""
],
[
"Keany",
"Jim",
""
],
[
"Joseph",
"Alan T",
""
],
[
"Kilambi",
"Srinivas",
""
]
] |
2505.22841 | Maria Han Veiga | Franck Gabriel, Fran\c{c}ois Ged, Maria Han Veiga and Emmanuel Schertzer | Kernel-Smoothed Scores for Denoising Diffusion: A Bias-Variance Study | null | null | null | null | cs.LG math.PR stat.ML | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Diffusion models now set the benchmark in high-fidelity generative sampling, yet they can, in principle, be prone to memorization. In this case, their learned score overfits the finite dataset so that the reverse-time SDE samples are mostly training points. In this paper, we interpret the empirical score as a noisy version of the true score and show that its covariance matrix is asymptotically a re-weighted data PCA. In large dimension, the small time limit makes the noise variance blow up while simultaneously reducing spatial correlation. To reduce this variance, we introduce a kernel-smoothed empirical score and analyze its bias-variance trade-off. We derive asymptotic bounds on the Kullback-Leibler divergence between the true distribution and the one generated by the modified reverse SDE. Regularization on the score has the same effect as increasing the size of the training dataset, and thus helps prevent memorization. A spectral decomposition of the forward diffusion suggests better variance control under some regularity conditions of the true data distribution. Reverse diffusion with kernel-smoothed empirical score can be reformulated as a gradient descent drifted toward a Log-Exponential Double-Kernel Density Estimator (LED-KDE). This perspective highlights two regularization mechanisms taking place in denoising diffusions: an initial Gaussian kernel first diffuses mass isotropically in the ambient space, while a second kernel applied in score space concentrates and spreads that mass along the data manifold. Hence, even a straightforward regularization-without any learning-already mitigates memorization and enhances generalization. Numerically, we illustrate our results with several experiments on synthetic and MNIST datasets. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:22:18 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Gabriel",
"Franck",
""
],
[
"Ged",
"François",
""
],
[
"Veiga",
"Maria Han",
""
],
[
"Schertzer",
"Emmanuel",
""
]
] |
2505.22842 | Lucas Kupssinsk\"u | Arthur S. Bianchessi, Rodrigo C. Barros, Lucas S. Kupssinsk\"u | Bayesian Attention Mechanism: A Probabilistic Framework for Positional Encoding and Context Length Extrapolation | null | null | null | null | cs.CL cs.LG | http://creativecommons.org/licenses/by/4.0/ | Transformer-based language models rely on positional encoding (PE) to handle token order and support context length extrapolation. However, existing PE methods lack theoretical clarity and rely on limited evaluation metrics to substantiate their extrapolation claims. We propose the Bayesian Attention Mechanism (BAM), a theoretical framework that formulates positional encoding as a prior within a probabilistic model. BAM unifies existing methods (e.g., NoPE and ALiBi) and motivates a new Generalized Gaussian positional prior that substantially improves long-context generalization. Empirically, BAM enables accurate information retrieval at $500\times$ the training context length, outperforming previous state-of-the-art context length generalization in long context retrieval accuracy while maintaining comparable perplexity and introducing minimal additional parameters. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:22:23 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Bianchessi",
"Arthur S.",
""
],
[
"Barros",
"Rodrigo C.",
""
],
[
"Kupssinskü",
"Lucas S.",
""
]
] |
2505.22843 | Aliai Eusebi | Alexander Herzog, Aliai Eusebi and Lorenzo Cavallaro | Aurora: Are Android Malware Classifiers Reliable under Distribution Shift? | null | null | null | null | cs.CR cs.AI | http://creativecommons.org/licenses/by/4.0/ | The performance figures of modern drift-adaptive malware classifiers appear promising, but does this translate to genuine operational reliability? The standard evaluation paradigm primarily focuses on baseline performance metrics, neglecting confidence-error alignment and operational stability. While TESSERACT established the importance of temporal evaluation, we take a complementary direction by investigating whether malware classifiers maintain reliable confidence estimates under distribution shifts and exploring the tensions between scientific advancement and practical impacts when they do not. We propose AURORA, a framework to evaluate malware classifiers based on their confidence quality and operational resilience. AURORA subjects the confidence profile of a given model to verification to assess the reliability of its estimates. Unreliable confidence estimates erode operational trust, waste valuable annotation budget on non-informative samples for active learning, and leave error-prone instances undetected in selective classification. AURORA is further complemented by a set of metrics designed to go beyond point-in-time performance, striving towards a more holistic assessment of operational stability throughout temporal evaluation periods. The fragility we observe in state-of-the-art frameworks across datasets of varying drift severity suggests the need for a return to the whiteboard. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:22:43 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Herzog",
"Alexander",
""
],
[
"Eusebi",
"Aliai",
""
],
[
"Cavallaro",
"Lorenzo",
""
]
] |
2505.22844 | Osama Bilal | Mahmoud M. Samak, Osama R. Bilal | Observation of dispersion anomalies by design | null | null | null | null | physics.app-ph | http://creativecommons.org/licenses/by/4.0/ | Band structures encode electronic, optical, and acoustic properties of matter and can serve as an essential tool in material discovery and design. Dispersion anomalies -- sharp, non-standard features in the frequency-wavenumber relation -- have been historically correlated with phonon-electron coupling or long-range interaction. Through a combination of experimental, numerical, and analytical methods, we show how magnetic couplings can induce negative stiffness and sculpt dispersion relations to support zero-frequency phonon anomalies at arbitrary, non-zero wavenumbers. Our approach enables the realization of complete wavenumber band gaps without time-modulation, electron-phonon coupling, or long-range interactions. We identify the conditions under which non-differentiable zero-frequency phonons exist away from the high-symmetry points. Our framework generalizes across monoatomic and diatomic lattices, locally resonant metamaterials, non-local systems, as well as higher dimensional crystals. In addition, we report the first passive- or active- experimental observation of wavenumber band gaps in higher dimensions. Our work establishes a new paradigm in dispersion engineering and provides means for understanding wave-matter interaction in both the frequency and wavenumber domains. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:22:54 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Samak",
"Mahmoud M.",
""
],
[
"Bilal",
"Osama R.",
""
]
] |
2505.22845 | Jonas Ricker | Sandra H\"oltervennhoff, Jonas Ricker, Maike M. Raphael, Charlotte Schwedes, Rebecca Weil, Asja Fischer, Thorsten Holz, Lea Sch\"onherr, Sascha Fahl | Security Benefits and Side Effects of Labeling AI-Generated Images | null | null | null | null | cs.CR cs.AI cs.CY cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Generative artificial intelligence is developing rapidly, impacting humans' interaction with information and digital media. It is increasingly used to create deceptively realistic misinformation, so lawmakers have imposed regulations requiring the disclosure of AI-generated content. However, only little is known about whether these labels reduce the risks of AI-generated misinformation.
Our work addresses this research gap. Focusing on AI-generated images, we study the implications of labels, including the possibility of mislabeling. Assuming that simplicity, transparency, and trust are likely to impact the successful adoption of such labels, we first qualitatively explore users' opinions and expectations of AI labeling using five focus groups. Second, we conduct a pre-registered online survey with over 1300 U.S. and EU participants to quantitatively assess the effect of AI labels on users' ability to recognize misinformation containing either human-made or AI-generated images. Our focus groups illustrate that, while participants have concerns about the practical implementation of labeling, they consider it helpful in identifying AI-generated images and avoiding deception. However, considering security benefits, our survey revealed an ambiguous picture, suggesting that users might over-rely on labels. While inaccurate claims supported by labeled AI-generated images were rated less credible than those with unlabeled AI-images, the belief in accurate claims also decreased when accompanied by a labeled AI-generated image. Moreover, we find the undesired side effect that human-made images conveying inaccurate claims were perceived as more credible in the presence of labels. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:24:45 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Höltervennhoff",
"Sandra",
""
],
[
"Ricker",
"Jonas",
""
],
[
"Raphael",
"Maike M.",
""
],
[
"Schwedes",
"Charlotte",
""
],
[
"Weil",
"Rebecca",
""
],
[
"Fischer",
"Asja",
""
],
[
"Holz",
"Thorsten",
""
],
[
"Schönherr",
"Lea",
""
],
[
"Fahl",
"Sascha",
""
]
] |
2505.22846 | Andrei Kozyrev | Nikita Khramov, Andrei Kozyrev, Gleb Solovev, Anton Podkopaev | RocqStar: Leveraging Similarity-driven Retrieval and Agentic Systems for Rocq generation | null | null | null | null | cs.LG cs.AI cs.LO cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Interactive Theorem Proving was repeatedly shown to be fruitful combined with Generative Artificial Intelligence. This paper assesses multiple approaches to Rocq generation and illuminates potential avenues for improvement. We highlight the importance of thorough premise selection for generating Rocq proofs and propose a novel approach, leveraging retrieval via a self-attentive embedder model. The evaluation of the designed approach shows up to 28% relative increase of the generator's performance. We tackle the problem of writing Rocq proofs using a multi-stage agentic system, tailored for formal verification, and demonstrate its high effectiveness. We conduct an ablation study and show the use of multi-agent debate on the planning stage of proof synthesis. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:26:11 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Khramov",
"Nikita",
""
],
[
"Kozyrev",
"Andrei",
""
],
[
"Solovev",
"Gleb",
""
],
[
"Podkopaev",
"Anton",
""
]
] |
2505.22847 | Clayton Shonkwiler | Mason Faldet and Clayton Shonkwiler | Sampling Finite Unit Norm Tight Frames Using Symplectic Geometry | 23 page, 5 figures | null | null | null | math.FA math.SG | http://creativecommons.org/licenses/by/4.0/ | Unit-norm tight frames in finite-dimensional Hilbert spaces (FUNTFs) are fundamental in signal processing, offering optimal robustness to noise and measurement loss. In this paper we introduce the Eigenlift algorithm for sampling random FUNTFs. Our approach exploits the symplectic geometry of the FUNTF space, which we characterize as a symplectic reduction of frame space by a symmetry group. We then define a Hamiltonian torus action on this reduced space whose momentum map induces a fiber bundle structure. The algorithm proceeds by sampling a point from the base space, which is a convex polytope, lifting it deterministically to a point on the corresponding fiber, then acting on this point by a random element of the torus to obtain a random FUNTF. We implement the method in Python and validate it in low-dimensional settings where it is computationally feasible to sample the base polytope via rejection sampling. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:29:03 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Faldet",
"Mason",
""
],
[
"Shonkwiler",
"Clayton",
""
]
] |
2505.22848 | Beiduo Chen | Pingjun Hong, Beiduo Chen, Siyao Peng, Marie-Catherine de Marneffe, Barbara Plank | LiTEx: A Linguistic Taxonomy of Explanations for Understanding Within-Label Variation in Natural Language Inference | 21 pages, 6 figures | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | There is increasing evidence of Human Label Variation (HLV) in Natural Language Inference (NLI), where annotators assign different labels to the same premise-hypothesis pair. However, within-label variation--cases where annotators agree on the same label but provide divergent reasoning--poses an additional and mostly overlooked challenge. Several NLI datasets contain highlighted words in the NLI item as explanations, but the same spans on the NLI item can be highlighted for different reasons, as evidenced by free-text explanations, which offer a window into annotators' reasoning. To systematically understand this problem and gain insight into the rationales behind NLI labels, we introduce LITEX, a linguistically-informed taxonomy for categorizing free-text explanations. Using this taxonomy, we annotate a subset of the e-SNLI dataset, validate the taxonomy's reliability, and analyze how it aligns with NLI labels, highlights, and explanations. We further assess the taxonomy's usefulness in explanation generation, demonstrating that conditioning generation on LITEX yields explanations that are linguistically closer to human explanations than those generated using only labels or highlights. Our approach thus not only captures within-label variation but also shows how taxonomy-guided generation for reasoning can bridge the gap between human and model explanations more effectively than existing strategies. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:32:48 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hong",
"Pingjun",
""
],
[
"Chen",
"Beiduo",
""
],
[
"Peng",
"Siyao",
""
],
[
"de Marneffe",
"Marie-Catherine",
""
],
[
"Plank",
"Barbara",
""
]
] |
2505.22849 | Dilara Aktas | Dilara Aktas, Ozgur B. Akan | Flexure-FET-Based Receiver with Competitive Binding for Interference Mitigation in Molecular Communication | null | null | null | null | eess.SP cs.SY eess.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Molecular communication (MC), a biologically inspired technology, enables applications in nanonetworks and the Internet of Everything (IoE), with great potential for intra-body systems such as drug delivery, health monitoring, and disease detection. This paper extends our prior work on the Flexure-FET MC receiver by integrating a competitive binding model to enhance performance in high-interference environments, where multiple molecular species coexist in the reception space. Previous studies have largely focused on ligand concentration estimation and detection, without fully addressing the effects of inter-species competition for receptor binding. Our proposed framework captures this competition, offering a more biologically accurate model for multitarget environments. By incorporating competition dynamics, the model improves understanding of MC behavior under interference. This approach enables fine-tuning of receptor responses by adjusting ligand concentrations and receptor affinities, thereby optimizing the performance of the Flexure-FET MC receiver. Comprehensive analysis shows that accounting for competitive binding is crucial for improving reliability and accuracy in complex MC systems. Factors such as signal-to-noise ratio (SNR), symbol error probability (SEP), interferer concentration, and receptor dynamics are shown to significantly affect performance. The proposed framework highlights the need to manage these factors effectively. Results demonstrate that modeling interference through competitive binding offers a realistic system perspective and allows tuning of receiver response, enabling robust detection in environments with multiple coexisting species. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:33:31 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Aktas",
"Dilara",
""
],
[
"Akan",
"Ozgur B.",
""
]
] |
2505.22850 | Panagiotis Kaliosis | Kostas Triaridis, Panagiotis Kaliosis, E-Ro Nguyen, Jingyi Xu, Hieu Le, Dimitris Samaras | Improving Contrastive Learning for Referring Expression Counting | 9 pages, 4 figures | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Object counting has progressed from class-specific models, which count only known categories, to class-agnostic models that generalize to unseen categories. The next challenge is Referring Expression Counting (REC), where the goal is to count objects based on fine-grained attributes and contextual differences. Existing methods struggle with distinguishing visually similar objects that belong to the same category but correspond to different referring expressions. To address this, we propose C-REX, a novel contrastive learning framework, based on supervised contrastive learning, designed to enhance discriminative representation learning. Unlike prior works, C-REX operates entirely within the image space, avoiding the misalignment issues of image-text contrastive learning, thus providing a more stable contrastive signal. It also guarantees a significantly larger pool of negative samples, leading to improved robustness in the learned representations. Moreover, we showcase that our framework is versatile and generic enough to be applied to other similar tasks like class-agnostic counting. To support our approach, we analyze the key components of sota detection-based models and identify that detecting object centroids instead of bounding boxes is the key common factor behind their success in counting tasks. We use this insight to design a simple yet effective detection-based baseline to build upon. Our experiments show that C-REX achieves state-of-the-art results in REC, outperforming previous methods by more than 22\% in MAE and more than 10\% in RMSE, while also demonstrating strong performance in class-agnostic counting. Code is available at https://github.com/cvlab-stonybrook/c-rex. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:33:42 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Triaridis",
"Kostas",
""
],
[
"Kaliosis",
"Panagiotis",
""
],
[
"Nguyen",
"E-Ro",
""
],
[
"Xu",
"Jingyi",
""
],
[
"Le",
"Hieu",
""
],
[
"Samaras",
"Dimitris",
""
]
] |
2505.22851 | Greg Muller | James Beyer, Jaewon Min, Greg Muller | Separating dots with circles | 18 pages, 8 figures | null | null | null | math.CO math.MG | http://creativecommons.org/licenses/by/4.0/ | Given a finite set of points in general position in the plane or sphere, we count the number of ways to separate those points using two types of circles: circles through three of the points, and circles through none of the points (up to an equivalence). In each case, we show the number of circles which separate the points into subsets of size k and l is independent of the configuration of points, and we provide an explicit formula in each case. We also consider how the circles change as the configuration of dots varies continuously. We show that an associated higher order Voronoi decomposition of the sphere changes by a sequence of local `moves'. As a consequence, an associated cluster algebra is independent of the configuration of dots, and only depends on the number of dots and the order of the Voronoi decomposition. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:34:56 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Beyer",
"James",
""
],
[
"Min",
"Jaewon",
""
],
[
"Muller",
"Greg",
""
]
] |
2505.22852 | Krti Tallam | Krti Tallam, Emma Miller | Operationalizing CaMeL: Strengthening LLM Defenses for Enterprise Deployment | null | null | null | null | cs.CR cs.AI | http://creativecommons.org/licenses/by/4.0/ | CaMeL (Capabilities for Machine Learning) introduces a capability-based sandbox to mitigate prompt injection attacks in large language model (LLM) agents. While effective, CaMeL assumes a trusted user prompt, omits side-channel concerns, and incurs performance tradeoffs due to its dual-LLM design. This response identifies these issues and proposes engineering improvements to expand CaMeL's threat coverage and operational usability. We introduce: (1) prompt screening for initial inputs, (2) output auditing to detect instruction leakage, (3) a tiered-risk access model to balance usability and control, and (4) a verified intermediate language for formal guarantees. Together, these upgrades align CaMeL with best practices in enterprise security and support scalable deployment. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:35:24 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Tallam",
"Krti",
""
],
[
"Miller",
"Emma",
""
]
] |
2505.22853 | Farrukh A. Chishtie | Farrukh A. Chishtie | A unified quaternion-complex framework for incompressible Navier-Stokes equations: new insights and implications | 15 pages, LaTeX | null | null | null | physics.flu-dyn math.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a novel, unified quaternion-complex framework for formulating the incompressible Navier-Stokes equations that reveals the geometric structure underlying viscous fluid motion and resolves the Clay Institute's Millennium Prize problem. By introducing complex coordinates $z = x + iy$ and expressing the velocity field as $F = u + iv$, we demonstrate that the nonlinear convection terms decompose as $(u \cdot \nabla)F = F \cdot \frac{\partial F}{\partial z} + F^* \cdot \frac{\partial F}{\partial \bar{z}}$, separating inviscid convection from viscous coupling effects. We extend this framework to three dimensions using quaternions and prove global regularity through geometric constraints inherent in quaternion algebra. The incompressibility constraint naturally emerges as a requirement that $\frac{\partial F}{\partial z}$ be purely imaginary, linking fluid mechanics to complex analysis fundamentally. Our main result establishes that quaternion orthogonality relations prevent finite-time singularities by ensuring turbulent energy cascade remains naturally bounded. The quaternion-complex formulation demonstrates that turbulence represents breakdown of quaternion-analyticity while maintaining geometric stability, providing rigorous mathematical foundation for understanding why real fluids exhibit finite turbulent behavior rather than mathematical singularities. We prove that for any smooth initial data, there exists a unique global smooth solution to the three-dimensional incompressible Navier-Stokes equations, directly resolving the Clay Institute challenge. Applications to atmospheric boundary layer physics demonstrate immediate practical relevance for environmental modeling, weather prediction, and climate modeling. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:37:33 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Chishtie",
"Farrukh A.",
""
]
] |
2505.22854 | Kornel Howil | Kornel Howil, Joanna Waczy\'nska, Piotr Borycki, Tadeusz Dziarmaga, Marcin Mazur, Przemys{\l}aw Spurek | CLIPGaussian: Universal and Multimodal Style Transfer Based on Gaussian Splatting | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Gaussian Splatting (GS) has recently emerged as an efficient representation for rendering 3D scenes from 2D images and has been extended to images, videos, and dynamic 4D content. However, applying style transfer to GS-based representations, especially beyond simple color changes, remains challenging. In this work, we introduce CLIPGaussians, the first unified style transfer framework that supports text- and image-guided stylization across multiple modalities: 2D images, videos, 3D objects, and 4D scenes. Our method operates directly on Gaussian primitives and integrates into existing GS pipelines as a plug-in module, without requiring large generative models or retraining from scratch. CLIPGaussians approach enables joint optimization of color and geometry in 3D and 4D settings, and achieves temporal coherence in videos, while preserving a model size. We demonstrate superior style fidelity and consistency across all tasks, validating CLIPGaussians as a universal and efficient solution for multimodal style transfer. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:41:24 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Howil",
"Kornel",
""
],
[
"Waczyńska",
"Joanna",
""
],
[
"Borycki",
"Piotr",
""
],
[
"Dziarmaga",
"Tadeusz",
""
],
[
"Mazur",
"Marcin",
""
],
[
"Spurek",
"Przemysław",
""
]
] |
2505.22855 | Ruining Deng | Ruining Deng, Junchao Zhu, Juming Xiong, Can Cui, Tianyuan Yao, Junlin Guo, Siqi Lu, Marilyn Lionts, Mengmeng Yin, Yu Wang, Shilin Zhao, Yucheng Tang, Yihe Yang, Paul Dennis Simonson, Mert R. Sabuncu, Haichun Yang, Yuankai Huo | IRS: Incremental Relationship-guided Segmentation for Digital Pathology | null | null | null | null | eess.IV cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Continual learning is rapidly emerging as a key focus in computer vision, aiming to develop AI systems capable of continuous improvement, thereby enhancing their value and practicality in diverse real-world applications. In healthcare, continual learning holds great promise for continuously acquired digital pathology data, which is collected in hospitals on a daily basis. However, panoramic segmentation on digital whole slide images (WSIs) presents significant challenges, as it is often infeasible to obtain comprehensive annotations for all potential objects, spanning from coarse structures (e.g., regions and unit objects) to fine structures (e.g., cells). This results in temporally and partially annotated data, posing a major challenge in developing a holistic segmentation framework. Moreover, an ideal segmentation model should incorporate new phenotypes, unseen diseases, and diverse populations, making this task even more complex. In this paper, we introduce a novel and unified Incremental Relationship-guided Segmentation (IRS) learning scheme to address temporally acquired, partially annotated data while maintaining out-of-distribution (OOD) continual learning capacity in digital pathology. The key innovation of IRS lies in its ability to realize a new spatial-temporal OOD continual learning paradigm by mathematically modeling anatomical relationships between existing and newly introduced classes through a simple incremental universal proposition matrix. Experimental results demonstrate that the IRS method effectively handles the multi-scale nature of pathological segmentation, enabling precise kidney segmentation across various structures (regions, units, and cells) as well as OOD disease lesions at multiple magnifications. This capability significantly enhances domain generalization, making IRS a robust approach for real-world digital pathology applications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:41:56 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Deng",
"Ruining",
""
],
[
"Zhu",
"Junchao",
""
],
[
"Xiong",
"Juming",
""
],
[
"Cui",
"Can",
""
],
[
"Yao",
"Tianyuan",
""
],
[
"Guo",
"Junlin",
""
],
[
"Lu",
"Siqi",
""
],
[
"Lionts",
"Marilyn",
""
],
[
"Yin",
"Mengmeng",
""
],
[
"Wang",
"Yu",
""
],
[
"Zhao",
"Shilin",
""
],
[
"Tang",
"Yucheng",
""
],
[
"Yang",
"Yihe",
""
],
[
"Simonson",
"Paul Dennis",
""
],
[
"Sabuncu",
"Mert R.",
""
],
[
"Yang",
"Haichun",
""
],
[
"Huo",
"Yuankai",
""
]
] |
2505.22856 | Jingcheng Zhou | Jingcheng Zhou, Senlei Li, Chuangtang Wang, Hanshang Jin, Stelo Xu, Zelong Xiong, Carson Jacobsen, Kenji Watanabe, Takashi Taniguchi, Valentin Taufour, Liuyan Zhao, Hua Chen, Chunhui Rita Du, Hailong Wang | Nanoscale quantum imaging of field-free deterministic switching of a chiral antiferromagnet | null | null | null | null | cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, unconventional spin-orbit torques (SOTs) with tunable spin generation open new pathways for designing novel magnetization control for cutting-edge spintronics innovations. A leading research thrust is to develop field-free deterministic magnetization switching for implementing scalable and energy favorable magnetic recording and storage applications, which have been demonstrated in conventional ferromagnetic and antiferromagnetic material systems. Here we extend this advanced magnetization control strategy to chiral antiferromagnet Mn3Sn using spin currents with out-of-plane canted polarization generated from low-symmetry van der Waals (vdW) material WTe2. Numerical calculations suggest that damping-like SOT of spins injected perpendicular to the kagome plane of Mn3Sn serves as a driving force to rotate the chiral magnetic order, while the field-like SOT of spin currents with polarization parallel to the kagome plane provides the bipolar deterministicity to the magnetic switching. We further introduce scanning quantum microscopy to visualize nanoscale evolutions of Mn3Sn magnetic domains during the field-free switching process, corroborating the exceptionally large magnetic switching ratio up to 90%. Our results highlight the opportunities provided by hybrid SOT material platforms consisting of noncollinear antiferromagnets and low-symmetry vdW spin source materials for developing next-generation, transformative spintronic logic devices. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:42:34 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhou",
"Jingcheng",
""
],
[
"Li",
"Senlei",
""
],
[
"Wang",
"Chuangtang",
""
],
[
"Jin",
"Hanshang",
""
],
[
"Xu",
"Stelo",
""
],
[
"Xiong",
"Zelong",
""
],
[
"Jacobsen",
"Carson",
""
],
[
"Watanabe",
"Kenji",
""
],
[
"Taniguchi",
"Takashi",
""
],
[
"Taufour",
"Valentin",
""
],
[
"Zhao",
"Liuyan",
""
],
[
"Chen",
"Hua",
""
],
[
"Du",
"Chunhui Rita",
""
],
[
"Wang",
"Hailong",
""
]
] |
2505.22857 | Vladimir Bataev | Vladimir Bataev, Andrei Andrusenko, Lilit Grigoryan, Aleksandr Laptev, Vitaly Lavrukhin, Boris Ginsburg | NGPU-LM: GPU-Accelerated N-Gram Language Model for Context-Biasing in Greedy ASR Decoding | Accepted to Interspeech 2025 | null | null | null | eess.AS cs.AI cs.CL cs.LG cs.SD | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Statistical n-gram language models are widely used for context-biasing tasks in Automatic Speech Recognition (ASR). However, existing implementations lack computational efficiency due to poor parallelization, making context-biasing less appealing for industrial use. This work rethinks data structures for statistical n-gram language models to enable fast and parallel operations for GPU-optimized inference. Our approach, named NGPU-LM, introduces customizable greedy decoding for all major ASR model types - including transducers, attention encoder-decoder models, and CTC - with less than 7% computational overhead. The proposed approach can eliminate more than 50% of the accuracy gap between greedy and beam search for out-of-domain scenarios while avoiding significant slowdown caused by beam search. The implementation of the proposed NGPU-LM is open-sourced. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:43:10 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Bataev",
"Vladimir",
""
],
[
"Andrusenko",
"Andrei",
""
],
[
"Grigoryan",
"Lilit",
""
],
[
"Laptev",
"Aleksandr",
""
],
[
"Lavrukhin",
"Vitaly",
""
],
[
"Ginsburg",
"Boris",
""
]
] |
2505.22858 | Sanjoy Kundu | Sanjoy Kundu, Shanmukha Vellamcheti and Sathyanarayanan N. Aakur | A Probabilistic Jump-Diffusion Framework for Open-World Egocentric Activity Recognition | Extended abstract of arXiv:2504.03948 for CVPR 2025 EgoVis Workshop | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Open-world egocentric activity recognition poses a fundamental challenge due to its unconstrained nature, requiring models to infer unseen activities from an expansive, partially observed search space. We introduce ProbRes, a Probabilistic Residual search framework based on jump-diffusion that efficiently navigates this space by balancing prior-guided exploration with likelihood-driven exploitation. Our approach integrates structured commonsense priors to construct a semantically coherent search space, adaptively refines predictions using Vision-Language Models (VLMs) and employs a stochastic search mechanism to locate high-likelihood activity labels while minimizing exhaustive enumeration efficiently. We systematically evaluate ProbRes across multiple openness levels (L0--L3), demonstrating its adaptability to increasing search space complexity. In addition to achieving state-of-the-art performance on benchmark datasets (GTEA Gaze, GTEA Gaze+, EPIC-Kitchens, and Charades-Ego), we establish a clear taxonomy for open-world recognition, delineating the challenges and methodological advancements necessary for egocentric activity understanding. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:44:14 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kundu",
"Sanjoy",
""
],
[
"Vellamcheti",
"Shanmukha",
""
],
[
"Aakur",
"Sathyanarayanan N.",
""
]
] |
2505.22859 | Hidenobu Matsuki | Hidenobu Matsuki, Gwangbin Bae and Andrew J. Davison | 4DTAM: Non-Rigid Tracking and Mapping via Dynamic Surface Gaussians | CVPR 2025. Project Page: https://muskie82.github.io/4dtam/ | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose the first 4D tracking and mapping method that jointly performs camera localization and non-rigid surface reconstruction via differentiable rendering. Our approach captures 4D scenes from an online stream of color images with depth measurements or predictions by jointly optimizing scene geometry, appearance, dynamics, and camera ego-motion. Although natural environments exhibit complex non-rigid motions, 4D-SLAM remains relatively underexplored due to its inherent challenges; even with 2.5D signals, the problem is ill-posed because of the high dimensionality of the optimization space. To overcome these challenges, we first introduce a SLAM method based on Gaussian surface primitives that leverages depth signals more effectively than 3D Gaussians, thereby achieving accurate surface reconstruction. To further model non-rigid deformations, we employ a warp-field represented by a multi-layer perceptron (MLP) and introduce a novel camera pose estimation technique along with surface regularization terms that facilitate spatio-temporal reconstruction. In addition to these algorithmic challenges, a significant hurdle in 4D SLAM research is the lack of reliable ground truth and evaluation protocols, primarily due to the difficulty of 4D capture using commodity sensors. To address this, we present a novel open synthetic dataset of everyday objects with diverse motions, leveraging large-scale object models and animation modeling. In summary, we open up the modern 4D-SLAM research by introducing a novel method and evaluation protocols grounded in modern vision and rendering techniques. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:45:10 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Matsuki",
"Hidenobu",
""
],
[
"Bae",
"Gwangbin",
""
],
[
"Davison",
"Andrew J.",
""
]
] |
2505.22860 | Bargav Jayaraman | Bargav Jayaraman, Virendra J. Marathe, Hamid Mozaffari, William F. Shen, Krishnaram Kenthapadi | Permissioned LLMs: Enforcing Access Control in Large Language Models | null | null | null | null | cs.CR cs.AI cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | In enterprise settings, organizational data is segregated, siloed and carefully protected by elaborate access control frameworks. These access control structures can completely break down if an LLM fine-tuned on the siloed data serves requests, for downstream tasks, from individuals with disparate access privileges. We propose Permissioned LLMs (PermLLM), a new class of LLMs that superimpose the organizational data access control structures on query responses they generate. We formalize abstractions underpinning the means to determine whether access control enforcement happens correctly over LLM query responses. Our formalism introduces the notion of a relevant response that can be used to prove whether a PermLLM mechanism has been implemented correctly. We also introduce a novel metric, called access advantage, to empirically evaluate the efficacy of a PermLLM mechanism. We introduce three novel PermLLM mechanisms that build on Parameter Efficient Fine-Tuning to achieve the desired access control. We furthermore present two instantiations of access advantage--(i) Domain Distinguishability Index (DDI) based on Membership Inference Attacks, and (ii) Utility Gap Index (UGI) based on LLM utility evaluation. We demonstrate the efficacy of our PermLLM mechanisms through extensive experiments on four public datasets (GPQA, RCV1, SimpleQA, and WMDP), in addition to evaluating the validity of DDI and UGI metrics themselves for quantifying access control in LLMs. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:47:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Jayaraman",
"Bargav",
""
],
[
"Marathe",
"Virendra J.",
""
],
[
"Mozaffari",
"Hamid",
""
],
[
"Shen",
"William F.",
""
],
[
"Kenthapadi",
"Krishnaram",
""
]
] |
2505.22861 | Carlota Pares Morlans | Carlota Par\'es-Morlans, Michelle Yi, Claire Chen, Sarah A. Wu, Rika Antonova, Tobias Gerstenberg, Jeannette Bohg | Causal-PIK: Causality-based Physical Reasoning with a Physics-Informed Kernel | Accepted to ICML 2025 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Tasks that involve complex interactions between objects with unknown dynamics make planning before execution difficult. These tasks require agents to iteratively improve their actions after actively exploring causes and effects in the environment. For these type of tasks, we propose Causal-PIK, a method that leverages Bayesian optimization to reason about causal interactions via a Physics-Informed Kernel to help guide efficient search for the best next action. Experimental results on Virtual Tools and PHYRE physical reasoning benchmarks show that Causal-PIK outperforms state-of-the-art results, requiring fewer actions to reach the goal. We also compare Causal-PIK to human studies, including results from a new user study we conducted on the PHYRE benchmark. We find that Causal-PIK remains competitive on tasks that are very challenging, even for human problem-solvers. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:51:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Parés-Morlans",
"Carlota",
""
],
[
"Yi",
"Michelle",
""
],
[
"Chen",
"Claire",
""
],
[
"Wu",
"Sarah A.",
""
],
[
"Antonova",
"Rika",
""
],
[
"Gerstenberg",
"Tobias",
""
],
[
"Bohg",
"Jeannette",
""
]
] |
2505.22862 | Andrew Choi | Yeon-Koo Che, Andrew B. Choi | Optimal Auction Design for Dynamic Stochastic Environments: Myerson Meets Naor | null | null | null | null | econ.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Allocation of goods and services often involves both stochastic supply and stochastic demand. Motivated by applications such as cloud computing, gig platforms, and blockchain auctions, we study the design of optimal selling mechanisms in an environment where buyers with private valuations arrive stochastically and are assigned goods that also arrive stochastically, and either buyers or goods can be held in a queue at costs until allocation. The optimal mechanism dynamically leverages competitive pressure across time by managing the queue of buyers and inventory of goods, using reserve prices that increase with the number of buyers in the queue and decrease with the number of items in inventory, and an auction to allocate the goods. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:51:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Che",
"Yeon-Koo",
""
],
[
"Choi",
"Andrew B.",
""
]
] |
2505.22863 | Yupei Li | Yupei Li, Shuaijie Shao, Manuel Milling, and Bj\"orn W. Schuller | Large Language Models for Depression Recognition in Spoken Language Integrating Psychological Knowledge | null | null | null | null | cs.HC cs.CL | http://creativecommons.org/licenses/by/4.0/ | Depression is a growing concern gaining attention in both public discourse and AI research. While deep neural networks (DNNs) have been used for recognition, they still lack real-world effectiveness. Large language models (LLMs) show strong potential but require domain-specific fine-tuning and struggle with non-textual cues. Since depression is often expressed through vocal tone and behaviour rather than explicit text, relying on language alone is insufficient. Diagnostic accuracy also suffers without incorporating psychological expertise. To address these limitations, we present, to the best of our knowledge, the first application of LLMs to multimodal depression detection using the DAIC-WOZ dataset. We extract the audio features using the pre-trained model Wav2Vec, and mapped it to text-based LLMs for further processing. We also propose a novel strategy for incorporating psychological knowledge into LLMs to enhance diagnostic performance, specifically using a question and answer set to grant authorised knowledge to LLMs. Our approach yields a notable improvement in both Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) compared to a base score proposed by the related original paper. The codes are available at https://github.com/myxp-lyp/Depression-detection.git | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:53:05 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Li",
"Yupei",
""
],
[
"Shao",
"Shuaijie",
""
],
[
"Milling",
"Manuel",
""
],
[
"Schuller",
"Björn W.",
""
]
] |
2505.22864 | Derek Weitzel | Derek Weitzel, Ashton Graves, Sam Albin, Huijun Zhu, Frank W\"urthwein, Mahidhar Tatineni, Dmitry Mishin, John Graham, Elham E Khoda, Mohammad Firas Sada, Larry Smarr, Thomas DeFanti | The National Research Platform: Stretched, Multi-Tenant, Scientific Kubernetes Cluster | Practice and Experience in Advanced Research Computing (PEARC '25), July 20--24, 2025, Columbus, OH, USA | null | 10.1145/3708035.3736060 | null | cs.DC | http://creativecommons.org/licenses/by/4.0/ | The National Research Platform (NRP) represents a distributed, multi-tenant Kubernetes-based cyberinfrastructure designed to facilitate collaborative scientific computing. Spanning over 75 locations in the U.S. and internationally, the NRP uniquely integrates varied computational resources, ranging from single nodes to extensive GPU and CPU clusters, to support diverse research workloads including advanced AI and machine learning tasks. It emphasizes flexibility through user-friendly interfaces such as JupyterHub and low level control of resources through direct Kubernetes interaction. Critical operational insights are discussed, including security enhancements using Kubernetes-integrated threat detection, extensive monitoring, and comprehensive accounting systems. This paper highlights the NRP's growing importance and scalability in addressing the increasing demands for distributed scientific computational resources. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:58:58 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Weitzel",
"Derek",
""
],
[
"Graves",
"Ashton",
""
],
[
"Albin",
"Sam",
""
],
[
"Zhu",
"Huijun",
""
],
[
"Würthwein",
"Frank",
""
],
[
"Tatineni",
"Mahidhar",
""
],
[
"Mishin",
"Dmitry",
""
],
[
"Graham",
"John",
""
],
[
"Khoda",
"Elham E",
""
],
[
"Sada",
"Mohammad Firas",
""
],
[
"Smarr",
"Larry",
""
],
[
"DeFanti",
"Thomas",
""
]
] |
2505.22865 | Susan Liang | Susan Liang, Dejan Markovic, Israel D. Gebru, Steven Krenn, Todd Keebler, Jacob Sandakly, Frank Yu, Samuel Hassel, Chenliang Xu, Alexander Richard | BinauralFlow: A Causal and Streamable Approach for High-Quality Binaural Speech Synthesis with Flow Matching Models | ICML 2025, 18 pages | null | null | null | cs.SD cs.AI eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Binaural rendering aims to synthesize binaural audio that mimics natural hearing based on a mono audio and the locations of the speaker and listener. Although many methods have been proposed to solve this problem, they struggle with rendering quality and streamable inference. Synthesizing high-quality binaural audio that is indistinguishable from real-world recordings requires precise modeling of binaural cues, room reverb, and ambient sounds. Additionally, real-world applications demand streaming inference. To address these challenges, we propose a flow matching based streaming binaural speech synthesis framework called BinauralFlow. We consider binaural rendering to be a generation problem rather than a regression problem and design a conditional flow matching model to render high-quality audio. Moreover, we design a causal U-Net architecture that estimates the current audio frame solely based on past information to tailor generative models for streaming inference. Finally, we introduce a continuous inference pipeline incorporating streaming STFT/ISTFT operations, a buffer bank, a midpoint solver, and an early skip schedule to improve rendering continuity and speed. Quantitative and qualitative evaluations demonstrate the superiority of our method over SOTA approaches. A perceptual study further reveals that our model is nearly indistinguishable from real-world recordings, with a $42\%$ confusion rate. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:59:15 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liang",
"Susan",
""
],
[
"Markovic",
"Dejan",
""
],
[
"Gebru",
"Israel D.",
""
],
[
"Krenn",
"Steven",
""
],
[
"Keebler",
"Todd",
""
],
[
"Sandakly",
"Jacob",
""
],
[
"Yu",
"Frank",
""
],
[
"Hassel",
"Samuel",
""
],
[
"Xu",
"Chenliang",
""
],
[
"Richard",
"Alexander",
""
]
] |
2505.22866 | Nicolas Espinosa Dice | Nicolas Espinosa-Dice, Yiyi Zhang, Yiding Chen, Bradley Guo, Owen Oertell, Gokul Swamy, Kiante Brantley, Wen Sun | Scaling Offline RL via Efficient and Expressive Shortcut Models | 32 pages, 5 figures. Under review at NeurIPS 2025 | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Diffusion and flow models have emerged as powerful generative approaches capable of modeling diverse and multimodal behavior. However, applying these models to offline reinforcement learning (RL) remains challenging due to the iterative nature of their noise sampling processes, making policy optimization difficult. In this paper, we introduce Scalable Offline Reinforcement Learning (SORL), a new offline RL algorithm that leverages shortcut models - a novel class of generative models - to scale both training and inference. SORL's policy can capture complex data distributions and can be trained simply and efficiently in a one-stage training procedure. At test time, SORL introduces both sequential and parallel inference scaling by using the learned Q-function as a verifier. We demonstrate that SORL achieves strong performance across a range of offline RL tasks and exhibits positive scaling behavior with increased test-time compute. We release the code at nico-espinosadice.github.io/projects/sorl. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:59:22 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Espinosa-Dice",
"Nicolas",
""
],
[
"Zhang",
"Yiyi",
""
],
[
"Chen",
"Yiding",
""
],
[
"Guo",
"Bradley",
""
],
[
"Oertell",
"Owen",
""
],
[
"Swamy",
"Gokul",
""
],
[
"Brantley",
"Kiante",
""
],
[
"Sun",
"Wen",
""
]
] |
2505.22867 | Kalina Bontcheva | Iknoor Singh, Carolina Scarton, Kalina Bontcheva | GateNLP at SemEval-2025 Task 10: Hierarchical Three-Step Prompting for Multilingual Narrative Classification | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The proliferation of online news and the increasing spread of misinformation necessitate robust methods for automatic data analysis. Narrative classification is emerging as a important task, since identifying what is being said online is critical for fact-checkers, policy markers and other professionals working on information studies. This paper presents our approach to SemEval 2025 Task 10 Subtask 2, which aims to classify news articles into a pre-defined two-level taxonomy of main narratives and sub-narratives across multiple languages.
We propose Hierarchical Three-Step Prompting (H3Prompt) for multilingual narrative classification. Our methodology follows a three-step Large Language Model (LLM) prompting strategy, where the model first categorises an article into one of two domains (Ukraine-Russia War or Climate Change), then identifies the most relevant main narratives, and finally assigns sub-narratives. Our approach secured the top position on the English test set among 28 competing teams worldwide. The code is available at https://github.com/GateNLP/H3Prompt. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 20:59:28 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Singh",
"Iknoor",
""
],
[
"Scarton",
"Carolina",
""
],
[
"Bontcheva",
"Kalina",
""
]
] |
2505.22868 | Md Hasibul Amin | Md Hasibul Amin, Mohammadreza Mohammadi, Jason D. Bakos, and Ramtin Zand | CrossNAS: A Cross-Layer Neural Architecture Search Framework for PIM Systems | null | null | null | null | cs.ET cs.AR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we propose the CrossNAS framework, an automated approach for exploring a vast, multidimensional search space that spans various design abstraction layers-circuits, architecture, and systems-to optimize the deployment of machine learning workloads on analog processing-in-memory (PIM) systems. CrossNAS leverages the single-path one-shot weight-sharing strategy combined with the evolutionary search for the first time in the context of PIM system mapping and optimization. CrossNAS sets a new benchmark for PIM neural architecture search (NAS), outperforming previous methods in both accuracy and energy efficiency while maintaining comparable or shorter search times. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 21:00:49 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Amin",
"Md Hasibul",
""
],
[
"Mohammadi",
"Mohammadreza",
""
],
[
"Bakos",
"Jason D.",
""
],
[
"Zand",
"Ramtin",
""
]
] |
2505.22869 | Junbo Yin | Junbo Yin, Chao Zha, Wenjia He, Chencheng Xu and Xin Gao | CFP-Gen: Combinatorial Functional Protein Generation via Diffusion Language Models | Accepted at ICML 2025. Code is available at https://github.com/yinjunbo/cfpgen | null | null | null | cs.CV cs.LG q-bio.BM | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Existing PLMs generate protein sequences based on a single-condition constraint from a specific modality, struggling to simultaneously satisfy multiple constraints across different modalities. In this work, we introduce CFP-Gen, a novel diffusion language model for Combinatorial Functional Protein GENeration. CFP-Gen facilitates the de novo protein design by integrating multimodal conditions with functional, sequence, and structural constraints. Specifically, an Annotation-Guided Feature Modulation (AGFM) module is introduced to dynamically adjust the protein feature distribution based on composable functional annotations, e.g., GO terms, IPR domains and EC numbers. Meanwhile, the Residue-Controlled Functional Encoding (RCFE) module captures residue-wise interaction to ensure more precise control. Additionally, off-the-shelf 3D structure encoders can be seamlessly integrated to impose geometric constraints. We demonstrate that CFP-Gen enables high-throughput generation of novel proteins with functionality comparable to natural proteins, while achieving a high success rate in designing multifunctional proteins. Code and data available at https://github.com/yinjunbo/cfpgen. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 21:05:46 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yin",
"Junbo",
""
],
[
"Zha",
"Chao",
""
],
[
"He",
"Wenjia",
""
],
[
"Xu",
"Chencheng",
""
],
[
"Gao",
"Xin",
""
]
] |
2505.22870 | Fernando G\'omez-Ortiz | Fernando G\'omez-Ortiz, Ramamoorthy Ramesh, Javier Junquera | Emergence of Transverse Dielectric Response in Ferroelectric Dielectric Heterostructures | null | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | We report the emergence of a transverse dielectric response in PbTiO$_{3}$/SrTiO$_{3}$ superlattices hosting polar vortex structures. Using second-principles simulations, we find that an electric field applied along one direction induces significant local polarization responses along orthogonal directions, with magnitudes approaching half that of the diagonal susceptibility components. These off-diagonal responses are strongly dependent on the topology of the vortex structure and can be deterministically tuned or even reversed via homogeneous electric fields or epitaxial strain. Notably, the transverse susceptibilities become comparable to the diagonal components during a field- or strain-induced transition to a polarization wave state. This discovery opens avenues for engineering reconfigurable nanoscale dielectric responses in topologically textured ferroelectric systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 21:07:17 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Gómez-Ortiz",
"Fernando",
""
],
[
"Ramesh",
"Ramamoorthy",
""
],
[
"Junquera",
"Javier",
""
]
] |
2505.22871 | Lior Limonad | Yuval David, Fabiana Fournier, Lior Limonad, Inna Skarbovsky | The WHY in Business Processes: Unification of Causal Process Models | 28 pages, 6 figures, BPM 2025 Forum | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Causal reasoning is essential for business process interventions and improvement, requiring a clear understanding of causal relationships among activity execution times in an event log. Recent work introduced a method for discovering causal process models but lacked the ability to capture alternating causal conditions across multiple variants. This raises the challenges of handling missing values and expressing the alternating conditions among log splits when blending traces with varying activities.
We propose a novel method to unify multiple causal process variants into a consistent model that preserves the correctness of the original causal models, while explicitly representing their causal-flow alternations. The method is formally defined, proved, evaluated on three open and two proprietary datasets, and released as an open-source implementation. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 21:12:30 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"David",
"Yuval",
""
],
[
"Fournier",
"Fabiana",
""
],
[
"Limonad",
"Lior",
""
],
[
"Skarbovsky",
"Inna",
""
]
] |
2505.22872 | Chao Liu | Chao Liu, Ankur Dhar, Ryan Herbst, Emilio A. Nanni | High Precision RF Pulse Shaping with Direct RF Sampling for Future Linear Accelerators | null | null | null | null | physics.acc-ph astro-ph.IM | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In various of particle accelerator designs, amplitude and phase modulation methods are commonly applied to shape the RF pulses for implementing pulse compressors or compensating for the fluctuations introduced by the high-power RF components and beam loading effects. Phase modulations are typically implemented with additional phase shifters that require drive or control electronics. With our recent next-generation LLRF (NG-LLRF) platform developed based on direct RF sampling technology of RF system-on-chip (RFSoC) devices, RF pulse shaping can be realized without the analogue phase shifters, which can significantly simplify the system architecture. We performed a range of high-power experiments in the C-band to evaluate the RF pulse-shaping capabilities of the NG-LLRF system at different stages of the RF circuits. In this paper, the high-power characterization results with the Cool Copper Collider (C3) structure driven by RF pulses with different modulation schemes will be described. With the pulse modulation and demodulation completely implemented in the digital domain, the RF pulse shaping schemes can be rapidly adapted for X-band structures simply by adding analogue mixers. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 21:13:35 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Liu",
"Chao",
""
],
[
"Dhar",
"Ankur",
""
],
[
"Herbst",
"Ryan",
""
],
[
"Nanni",
"Emilio A.",
""
]
] |
2505.22873 | Stephen Lee J | Stephen J. Lee and Cailinn Drouin | Forecasting Residential Heating and Electricity Demand with Scalable, High-Resolution, Open-Source Models | 16 pages, 4 figures, 1 table, preprint | null | null | null | econ.GN q-fin.EC stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a novel framework for high-resolution forecasting of residential heating and electricity demand using probabilistic deep learning models. We focus specifically on providing hourly building-level electricity and heating demand forecasts for the residential sector. Leveraging multimodal building-level information -- including data on building footprint areas, heights, nearby building density, nearby building size, land use patterns, and high-resolution weather data -- and probabilistic modeling, our methods provide granular insights into demand heterogeneity. Validation at the building level underscores a step change improvement in performance relative to NREL's ResStock model, which has emerged as a research community standard for residential heating and electricity demand characterization. In building-level heating and electricity estimation backtests, our probabilistic models respectively achieve RMSE scores 18.3\% and 35.1\% lower than those based on ResStock. By offering an open-source, scalable, high-resolution platform for demand estimation and forecasting, this research advances the tools available for policymakers and grid planners, contributing to the broader effort to decarbonize the U.S. building stock and meeting climate objectives. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 21:16:27 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lee",
"Stephen J.",
""
],
[
"Drouin",
"Cailinn",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.