# Research Papers — 2025-12 · bundle page 139 of 175 > Full-text markdown of papers #6901–6950 of 8703 published in 2025-12, newest first. AI-readable mirrors of the full 8703-paper 2025-12 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2025-12/page/139 [← Previous bundle](https://aipod.dev/papers/archive/2025-12/page/138.md) [Next bundle →](https://aipod.dev/papers/archive/2025-12/page/140.md) ### SIMPACT: Simulation-Enabled Action Planning using Vision-Language Models Authors: Haowen Liu, Shaoxiong Yao, Haonan Chen, Jiawei Gao, Jiayuan Mao, Jia-Bin Huang, Yilun Du Published: 2025-12-05 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2512.05955v2 Vision-Language Models (VLMs) exhibit remarkable common-sense and semantic reasoning capabilities. However, they lack a grounded understanding of physical dynamics. This limitation arises from training VLMs on static internet-scale visual-language data that contain no causal interactions or action-conditioned changes. Consequently, it remains challenging to leverage VLMs for fine-grained robotic manipulation tasks that require physical understanding, reasoning, and corresponding action planning. To overcome this, we present SIMPACT, a test-time, SIMulation-enabled ACTion Planning framework that equips VLMs with physical reasoning through simulation-in-the-loop world modeling, without requiring any additional training. From a single RGB-D observation, SIMPACT efficiently constructs physics simulations, enabling the VLM to propose informed actions, observe simulated rollouts, and iteratively refine its reasoning. By integrating language reasoning with physics prediction, our simulation-enabled VLM can understand contact dynamics and action outcomes in a physically grounded way. Our method demonstrates state-of-the-art performance on five challenging, real-world rigid-body and deformable manipulation tasks that require fine-grained physical reasoning, outperforming existing general-purpose robotic manipulation models. Our results demonstrate that embedding physics understanding via efficient simulation into VLM reasoning at test time offers a promising path towards generalizable embodied intelligence. Project webpage can be found at https://simpact-bot.github.io --- ### SIMPACT: Simulation-Enabled Action Planning using Vision-Language Models Authors: Haowen Liu, Shaoxiong Yao, Haonan Chen, Jiawei Gao, Jiayuan Mao, Jia-Bin Huang, Yilun Du Published: 2025-12-05 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2512.05955v1 Vision-Language Models (VLMs) exhibit remarkable common-sense and semantic reasoning capabilities. However, they lack a grounded understanding of physical dynamics. This limitation arises from training VLMs on static internet-scale visual-language data that contain no causal interactions or action-conditioned changes. Consequently, it remains challenging to leverage VLMs for fine-grained robotic manipulation tasks that require physical understanding, reasoning, and corresponding action planning. To overcome this, we present SIMPACT, a test-time, SIMulation-enabled ACTion Planning framework that equips VLMs with physical reasoning through simulation-in-the-loop world modeling, without requiring any additional training. From a single RGB-D observation, SIMPACT efficiently constructs physics simulations, enabling the VLM to propose informed actions, observe simulated rollouts, and iteratively refine its reasoning. By integrating language reasoning with physics prediction, our simulation-enabled VLM can understand contact dynamics and action outcomes in a physically grounded way. Our method demonstrates state-of-the-art performance on five challenging, real-world rigid-body and deformable manipulation tasks that require fine-grained physical reasoning, outperforming existing general-purpose robotic manipulation models. Our results demonstrate that embedding physics understanding via efficient simulation into VLM reasoning at test time offers a promising path towards generalizable embodied intelligence. Project webpage can be found at https://simpact-bot.github.io --- ### SymPyBench: A Dynamic Benchmark for Scientific Reasoning with Executable Python Code Authors: Shima Imani, Seungwhan Moon, Adel Ahmadyan, Lu Zhang, Kirmani Ahmed, Babak Damavandi Published: 2025-12-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.05954v1 We introduce, a large-scale synthetic benchmark of 15,045 university-level physics problems (90/10% train/test split). Each problem is fully parameterized, supporting an effectively infinite range of input configurations, and is accompanied by structured, step-by-step reasoning and executable Python code that produces the ground-truth solution for any parameter set. The benchmark contains three question types: MC-Symbolic (multiple-choice with symbolic options), MC-Numerical (multiple-choice with numerical options), and free-form (open-ended responses). These diverse formats test complementary reasoning skills. By leveraging the dynamic, code-driven nature of the benchmark, we introduce three novel evaluation metrics in addition to standard accuracy: Consistency Score, Failure Rate, and Confusion Rate, that quantify variability and uncertainty across problem variants. Experiments with state-of-the-art instruction-tuned language models reveal both strengths and limitations in scientific reasoning, positioning SymPyBench as a foundation for developing more robust and interpretable reasoning systems --- ### Trusted AI Agents in the Cloud Authors: Teofil Bodea, Masanori Misono, Julian Pritzi, Patrick Sabanic, Thore Sommer, Harshavardhan Unnibhavi, David Schall, Nuno Santos, Dimitrios Stavrakakis, Pramod Bhatotia Published: 2025-12-05 Categories: cs.CR, cs.AI, cs.MA arXiv: https://arxiv.org/abs/2512.05951v2 AI agents powered by large language models are increasingly deployed as cloud services that autonomously access sensitive data, invoke external tools, and interact with other agents. However, these agents run within a complex multi-party ecosystem, where untrusted components can lead to data leakage, tampering, or unintended behavior. Existing Confidential Virtual Machines (CVMs) provide only per binary protection and offer no guarantees for cross-principal trust, accelerator-level isolation, or supervised agent behavior. We present Omega, a system that enables trusted AI agents by enforcing end-to-end isolation, establishing verifiable trust across all contributing principals, and supervising every external interaction with accountable provenance. Omega builds on Confidential VMs and Confidential GPUs to create a Trusted Agent Platform that hosts many agents within a single CVM using nested isolation. It also provides efficient multi-agent orchestration with cross-principal trust establishment via differential attestation, and a policy specification and enforcement framework that governs data access, tool usage, and inter-agent communication for data protection and regulatory compliance. Implemented on AMD SEV-SNP and NVIDIA H100, Omega fully secures agent state across CVM-GPU, and achieves high performance while enabling high-density, policy-compliant multi-agent deployments at cloud scale. --- ### Trusted AI Agents in the Cloud Authors: Teofil Bodea, Masanori Misono, Julian Pritzi, Patrick Sabanic, Thore Sommer, Harshavardhan Unnibhavi, David Schall, Nuno Santos, Dimitrios Stavrakakis, Pramod Bhatotia Published: 2025-12-05 Categories: cs.CR, cs.AI, cs.MA arXiv: https://arxiv.org/abs/2512.05951v1 AI agents powered by large language models are increasingly deployed as cloud services that autonomously access sensitive data, invoke external tools, and interact with other agents. However, these agents run within a complex multi-party ecosystem, where untrusted components can lead to data leakage, tampering, or unintended behavior. Existing Confidential Virtual Machines (CVMs) provide only per binary protection and offer no guarantees for cross-principal trust, accelerator-level isolation, or supervised agent behavior. We present Omega, a system that enables trusted AI agents by enforcing end-to-end isolation, establishing verifiable trust across all contributing principals, and supervising every external interaction with accountable provenance. Omega builds on Confidential VMs and Confidential GPUs to create a Trusted Agent Platform that hosts many agents within a single CVM using nested isolation. It also provides efficient multi-agent orchestration with cross-principal trust establishment via differential attestation, and a policy specification and enforcement framework that governs data access, tool usage, and inter-agent communication for data protection and regulatory compliance. Implemented on AMD SEV-SNP and NVIDIA H100, Omega fully secures agent state across CVM-GPU, and achieves high performance while enabling high-density, policy-compliant multi-agent deployments at cloud scale. --- ### Impugan: Learning Conditional Generative Models for Robust Data Imputation Authors: Zalish Mahmud, Anantaa Kotal, Aritran Piplai Published: 2025-12-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.05950v1 Incomplete data are common in real-world applications. Sensors fail, records are inconsistent, and datasets collected from different sources often differ in scale, sampling rate, and quality. These differences create missing values that make it difficult to combine data and build reliable models. Standard imputation methods such as regression models, expectation-maximization, and multiple imputation rely on strong assumptions about linearity and independence. These assumptions rarely hold for complex or heterogeneous data, which can lead to biased or over-smoothed estimates. We propose Impugan, a conditional Generative Adversarial Network (cGAN) for imputing missing values and integrating heterogeneous datasets. The model is trained on complete samples to learn how missing variables depend on observed ones. During inference, the generator reconstructs missing entries from available features, and the discriminator enforces realism by distinguishing true from imputed data. This adversarial process allows Impugan to capture nonlinear and multimodal relationships that conventional methods cannot represent. In experiments on benchmark datasets and a multi-source integration task, Impugan achieves up to 82\% lower Earth Mover's Distance (EMD) and 70\% lower mutual-information deviation (MI) compared to leading baselines. These results show that adversarially trained generative models provide a scalable and principled approach for imputing and merging incomplete, heterogeneous data. Our model is available at: github.com/zalishmahmud/impuganBigData2025 --- ### Developing synthetic microdata through machine learning for firm-level business surveys Authors: Jorge Cisneros, Timothy Wojan, Matthew Williams, Jennifer Ozawa, Robert Chew, Kimberly Janda, Timothy Navarro, Michael Floyd, Christine Task, Damon Streat Published: 2025-12-05 Categories: cs.LG, econ.GN, stat.AP, stat.ME arXiv: https://arxiv.org/abs/2512.05948v2 Public-use microdata samples (PUMS) from the United States (US) Census Bureau on individuals have been available for decades. However, large increases in computing power and the greater availability of Big Data have dramatically increased the probability of re-identifying anonymized data, potentially violating the pledge of confidentiality given to survey respondents. Data science tools can be used to produce synthetic data that preserve critical moments of the empirical data but do not contain the records of any existing individual respondent or business. Developing public-use firm data from surveys presents unique challenges different from demographic data, because there is a lack of anonymity and certain industries can be easily identified in each geographic area. This paper briefly describes a machine learning model used to construct a synthetic PUMS based on the Annual Business Survey (ABS) and discusses various quality metrics. Although the ABS PUMS is currently being refined and results are confidential, we present two synthetic PUMS developed for the 2007 Survey of Business Owners, similar to the ABS business data. Econometric replication of a high impact analysis published in Small Business Economics demonstrates the verisimilitude of the synthetic data to the true data and motivates discussion of possible ABS use cases. --- ### Developing synthetic microdata through machine learning for firm-level business surveys Authors: Jorge Cisneros Paz, Timothy Wojan, Matthew Williams, Jennifer Ozawa, Robert Chew, Kimberly Janda, Timothy Navarro, Michael Floyd, Christine Task, Damon Streat Published: 2025-12-05 Categories: cs.LG, econ.GN, stat.AP, stat.ME arXiv: https://arxiv.org/abs/2512.05948v1 Public-use microdata samples (PUMS) from the United States (US) Census Bureau on individuals have been available for decades. However, large increases in computing power and the greater availability of Big Data have dramatically increased the probability of re-identifying anonymized data, potentially violating the pledge of confidentiality given to survey respondents. Data science tools can be used to produce synthetic data that preserve critical moments of the empirical data but do not contain the records of any existing individual respondent or business. Developing public-use firm data from surveys presents unique challenges different from demographic data, because there is a lack of anonymity and certain industries can be easily identified in each geographic area. This paper briefly describes a machine learning model used to construct a synthetic PUMS based on the Annual Business Survey (ABS) and discusses various quality metrics. Although the ABS PUMS is currently being refined and results are confidential, we present two synthetic PUMS developed for the 2007 Survey of Business Owners, similar to the ABS business data. Econometric replication of a high impact analysis published in Small Business Economics demonstrates the verisimilitude of the synthetic data to the true data and motivates discussion of possible ABS use cases. --- ### Variational Quantum Rainbow Deep Q-Network for Optimizing Resource Allocation Problem Authors: Truong Thanh Hung Nguyen, Truong Thinh Nguyen, Hung Cao Published: 2025-12-05 Categories: cs.AI, cs.ET, cs.SE arXiv: https://arxiv.org/abs/2512.05946v1 Resource allocation remains NP-hard due to combinatorial complexity. While deep reinforcement learning (DRL) methods, such as the Rainbow Deep Q-Network (DQN), improve scalability through prioritized replay and distributional heads, classical function approximators limit their representational power. We introduce Variational Quantum Rainbow DQN (VQR-DQN), which integrates ring-topology variational quantum circuits with Rainbow DQN to leverage quantum superposition and entanglement. We frame the human resource allocation problem (HRAP) as a Markov decision process (MDP) with combinatorial action spaces based on officer capabilities, event schedules, and transition times. On four HRAP benchmarks, VQR-DQN achieves 26.8% normalized makespan reduction versus random baselines and outperforms Double DQN and classical Rainbow DQN by 4.9-13.4%. These gains align with theoretical connections between circuit expressibility, entanglement, and policy quality, demonstrating the potential of quantum-enhanced DRL for large-scale resource allocation. Our implementation is available at: https://github.com/Analytics-Everywhere-Lab/qtrl/. --- ### TRACE: A Framework for Analyzing and Enhancing Stepwise Reasoning in Vision-Language Models Authors: Shima Imani, Seungwhan Moon, Lambert Mathias, Lu Zhang, Babak Damavandi Published: 2025-12-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.05943v3 Reliable mathematical and scientific reasoning remains an open challenge for large vision-language models. Standard final-answer evaluation often masks reasoning errors, allowing silent failures to persist. To address this gap, we introduce TRACE, a framework for Transparent Reasoning And Consistency Evaluation that diagnoses reasoning trajectories rather than only end results. At its core, TRACE leverages Auxiliary Reasoning Sets, compact sub question answer pairs that decompose complex problems, evaluate intermediate steps through consistency-based metrics, and expose failures overlooked by standard evaluation. Our experiments show that consistency across ARS correlates with final-answer correctness and helps pinpoint the reasoning steps where failures arise, offering actionable signals for model improvement. Furthermore, TRACE defines confidence regions that distinguish reliable from unreliable reasoning paths, supporting effective filtering, debugging, and model refinement. --- ### TRACE: A Framework for Analyzing and Enhancing Stepwise Reasoning in Vision-Language Models Authors: Shima Imani, Seungwhan Moon, Lambert Mathias, Lu Zhang, Babak Damavandi Published: 2025-12-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.05943v1 Reliable mathematical and scientific reasoning remains an open challenge for large vision-language models. Standard final-answer evaluation often masks reasoning errors, allowing silent failures to persist. To address this gap, we introduce TRACE, a framework for Transparent Reasoning And Consistency Evaluation that diagnoses reasoning trajectories rather than only end results. At its core, TRACE leverages Auxiliary Reasoning Sets, compact sub question answer pairs that decompose complex problems, evaluate intermediate steps through consistency-based metrics, and expose failures overlooked by standard evaluation. Our experiments show that consistency across ARS correlates with final-answer correctness and helps pinpoint the reasoning steps where failures arise, offering actionable signals for model improvement. Furthermore, TRACE defines confidence regions that distinguish reliable from unreliable reasoning paths, supporting effective filtering, debugging, and model refinement. --- ### Zoom in, Click out: Unlocking and Evaluating the Potential of Zooming for GUI Grounding Authors: Zhiyuan Jiang, Shenghao Xie, Wenyi Li, Wenqiang Zu, Peihang Li, Jiahao Qiu, Siqi Pei, Lei Ma, Tiejun Huang, Mengdi Wang, Shilong Liu Published: 2025-12-05 Categories: cs.CV, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2512.05941v1 Grounding is a fundamental capability for building graphical user interface (GUI) agents. Although existing approaches rely on large-scale bounding box supervision, they still face various challenges, such as cross-platform generalization, complex layout analysis, and fine-grained element localization. In this paper, we investigate zoom as a strong yet underexplored prior for GUI grounding, and propose a training-free method, ZoomClick. By characterizing four key properties of zoom (i.e., pre-zoom, depth, shrink size, minimal crop size), we unlock its full capabilities for dynamic spatial focusing and adaptive context switching. Experiments demonstrate that our method significantly boosts the performance of both general vision-language and specialized GUI grounding models, achieving state-of-the-art results on several mainstream benchmarks; for example, UI-Venus-72B attains a 73.1% success rate on ScreenSpot-Pro. Furthermore, we present GUIZoom-Bench, a benchmark for evaluating model adaptability to zoom, aiming to inspire future research on improving zoom for further training and test-time scaling in GUI grounding tasks. --- ### Designing an Optimal Sensor Network via Minimizing Information Loss Authors: Daniel Waxman, Fernando Llorente, Katia Lamer, Petar M. Djurić Published: 2025-12-05 Categories: stat.ME, cs.LG, stat.CO, stat.ML arXiv: https://arxiv.org/abs/2512.05940v2 Optimal experimental design is a classic topic in statistics, with many well-studied problems, applications, and solutions. The design problem we study is the placement of sensors to monitor spatiotemporal processes, explicitly accounting for the temporal dimension in our modeling and optimization. We observe that recent advancements in computational sciences often yield large datasets based on physics-based simulations, which are rarely leveraged in experimental design. We introduce a novel model-based sensor placement criterion, along with a highly-efficient optimization algorithm, which integrates physics-based simulations and Bayesian experimental design principles to identify sensor networks that "minimize information loss" from simulated data. Our technique relies on sparse variational inference and (separable) Gauss-Markov priors, and thus may adapt many techniques from Bayesian experimental design. We validate our method through a case study monitoring air temperature in Phoenix, Arizona, using state-of-the-art physics-based simulations. Our results show our framework to be superior to random or quasi-random sampling, particularly with a limited number of sensors. We conclude by discussing practical considerations and implications of our framework, including more complex modeling tools and real-world deployments. --- ### Designing an Optimal Sensor Network via Minimizing Information Loss Authors: Daniel Waxman, Fernando Llorente, Katia Lamer, Petar M. Djurić Published: 2025-12-05 Categories: stat.ME, cs.LG, stat.CO, stat.ML arXiv: https://arxiv.org/abs/2512.05940v1 Optimal experimental design is a classic topic in statistics, with many well-studied problems, applications, and solutions. The design problem we study is the placement of sensors to monitor spatiotemporal processes, explicitly accounting for the temporal dimension in our modeling and optimization. We observe that recent advancements in computational sciences often yield large datasets based on physics-based simulations, which are rarely leveraged in experimental design. We introduce a novel model-based sensor placement criterion, along with a highly-efficient optimization algorithm, which integrates physics-based simulations and Bayesian experimental design principles to identify sensor networks that "minimize information loss" from simulated data. Our technique relies on sparse variational inference and (separable) Gauss-Markov priors, and thus may adapt many techniques from Bayesian experimental design. We validate our method through a case study monitoring air temperature in Phoenix, Arizona, using state-of-the-art physics-based simulations. Our results show our framework to be superior to random or quasi-random sampling, particularly with a limited number of sensors. We conclude by discussing practical considerations and implications of our framework, including more complex modeling tools and real-world deployments. --- ### Measuring the Effect of Background on Classification and Feature Importance in Deep Learning for AV Perception Authors: Anne Sielemann, Valentin Barner, Stefan Wolf, Masoud Roschani, Jens Ziehn, Juergen Beyerer Published: 2025-12-05 Categories: cs.CV, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2512.05937v1 Common approaches to explainable AI (XAI) for deep learning focus on analyzing the importance of input features on the classification task in a given model: saliency methods like SHAP and GradCAM are used to measure the impact of spatial regions of the input image on the classification result. Combined with ground truth information about the location of the object in the input image (e.g., a binary mask), it is determined whether object pixels had a high impact on the classification result, or whether the classification focused on background pixels. The former is considered to be a sign of a healthy classifier, whereas the latter is assumed to suggest overfitting on spurious correlations. A major challenge, however, is that these intuitive interpretations are difficult to test quantitatively, and hence the output of such explanations lacks an explanation itself. One particular reason is that correlations in real-world data are difficult to avoid, and whether they are spurious or legitimate is debatable. Synthetic data in turn can facilitate to actively enable or disable correlations where desired but often lack a sufficient quantification of realism and stochastic properties. (...) Therefore, we systematically generate six synthetic datasets for the task of traffic sign recognition, which differ only in their degree of camera variation and background correlation (...) to quantify the isolated influence of background correlation, different levels of camera variation, and considered traffic sign shapes on the classification performance, as well as background feature importance. (...) Results include a quantification of when and how much background features gain importance to support the classification task based on changes in the training domain (...). Download: synset.de/datasets/synset-signset-ger/background-effect --- ### Synset Signset Germany: a Synthetic Dataset for German Traffic Sign Recognition Authors: Anne Sielemann, Lena Loercher, Max-Lion Schumacher, Stefan Wolf, Masoud Roschani, Jens Ziehn Published: 2025-12-05 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2512.05936v1 In this paper, we present a synthesis pipeline and dataset for training / testing data in the task of traffic sign recognition that combines the advantages of data-driven and analytical modeling: GAN-based texture generation enables data-driven dirt and wear artifacts, rendering unique and realistic traffic sign surfaces, while the analytical scene modulation achieves physically correct lighting and allows detailed parameterization. In particular, the latter opens up applications in the context of explainable AI (XAI) and robustness tests due to the possibility of evaluating the sensitivity to parameter changes, which we demonstrate with experiments. Our resulting synthetic traffic sign recognition dataset Synset Signset Germany contains a total of 105500 images of 211 different German traffic sign classes, including newly published (2020) and thus comparatively rare traffic signs. In addition to a mask and a segmentation image, we also provide extensive metadata including the stochastically selected environment and imaging effect parameters for each image. We evaluate the degree of realism of Synset Signset Germany on the real-world German Traffic Sign Recognition Benchmark (GTSRB) and in comparison to CATERED, a state-of-the-art synthetic traffic sign recognition dataset. --- ### Physically-Based Simulation of Automotive LiDAR Authors: L. Dudzik, M. Roschani, A. Sielemann, K. Trampert, J. Ziehn, J. Beyerer, C. Neumann Published: 2025-12-05 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2512.05932v1 We present an analytic model for simulating automotive time-of-flight (ToF) LiDAR that includes blooming, echo pulse width, and ambient light, along with steps to determine model parameters systematically through optical laboratory measurements. The model uses physically based rendering (PBR) in the near-infrared domain. It assumes single-bounce reflections and retroreflections over rasterized rendered images from shading or ray tracing, including light emitted from the sensor as well as stray light from other, non-correlated sources such as sunlight. Beams from the sensor and sensitivity of the receiving diodes are modeled with flexible beam steering patterns and with non-vanishing diameter. Different (all non-real time) computational approaches can be chosen based on system properties, computing capabilities, and desired output properties. Model parameters include system-specific properties, namely the physical spread of the LiDAR beam, combined with the sensitivity of the receiving diode; the intensity of the emitted light; the conversion between the intensity of reflected light and the echo pulse width; and scenario parameters such as environment lighting, positioning, and surface properties of the target(s) in the relevant infrared domain. System-specific properties of the model are determined from laboratory measurements of the photometric luminance on different target surfaces aligned with a goniometer at 0.01° resolution, which marks the best available resolution for measuring the beam pattern. The approach is calibrated for and tested on two automotive LiDAR systems, the Valeo Scala Gen. 2 and the Blickfeld Cube 1. Both systems differ notably in their properties and available interfaces, but the relevant model parameters could be extracted successfully. --- ### On the Bayes Inconsistency of Disagreement Discrepancy Surrogates Authors: Neil G. Marchant, Andrew C. Cullen, Feng Liu, Sarah M. Erfani Published: 2025-12-05 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2512.05931v1 Deep neural networks often fail when deployed in real-world contexts due to distribution shift, a critical barrier to building safe and reliable systems. An emerging approach to address this problem relies on \emph{disagreement discrepancy} -- a measure of how the disagreement between two models changes under a shifting distribution. The process of maximizing this measure has seen applications in bounding error under shifts, testing for harmful shifts, and training more robust models. However, this optimization involves the non-differentiable zero-one loss, necessitating the use of practical surrogate losses. We prove that existing surrogates for disagreement discrepancy are not Bayes consistent, revealing a fundamental flaw: maximizing these surrogates can fail to maximize the true disagreement discrepancy. To address this, we introduce new theoretical results providing both upper and lower bounds on the optimality gap for such surrogates. Guided by this theory, we propose a novel disagreement loss that, when paired with cross-entropy, yields a provably consistent surrogate for disagreement discrepancy. Empirical evaluations across diverse benchmarks demonstrate that our method provides more accurate and robust estimates of disagreement discrepancy than existing approaches, particularly under challenging adversarial conditions. --- ### PRiSM: An Agentic Multimodal Benchmark for Scientific Reasoning via Python-Grounded Evaluation Authors: Shima Imani, Seungwhan Moon, Adel Ahmadyan, Lu Zhang, Kirmani Ahmed, Babak Damavandi Published: 2025-12-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.05930v1 Evaluating vision-language models (VLMs) in scientific domains like mathematics and physics poses unique challenges that go far beyond predicting final answers. These domains demand conceptual understanding, symbolic reasoning, and adherence to formal laws, requirements that most existing benchmarks fail to address. In particular, current datasets tend to be static, lacking intermediate reasoning steps, robustness to variations, or mechanisms for verifying scientific correctness. To address these limitations, we introduce PRiSM, a synthetic, fully dynamic, and multimodal benchmark for evaluating scientific reasoning via grounded Python code. PRiSM includes over 24,750 university-level physics and math problems, and it leverages our scalable agent-based pipeline, PrismAgent, to generate well-structured problem instances. Each problem contains dynamic textual and visual input, a generated figure, alongside rich structured outputs: executable Python code for ground truth generation and verification, and detailed step-by-step reasoning. The dynamic nature and Python-powered automated ground truth generation of our benchmark allow for fine-grained experimental auditing of multimodal VLMs, revealing failure modes, uncertainty behaviors, and limitations in scientific reasoning. To this end, we propose five targeted evaluation tasks covering generalization, symbolic program synthesis, perturbation robustness, reasoning correction, and ambiguity resolution. Through comprehensive evaluation of existing VLMs, we highlight their limitations and showcase how PRiSM enables deeper insights into their scientific reasoning capabilities. --- ### A Comparative Study on Synthetic Facial Data Generation Techniques for Face Recognition Authors: Pedro Vidal, Bernardo Biesseck, Luiz E. L. Coelho, Roger Granada, David Menotti Published: 2025-12-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.05928v1 Facial recognition has become a widely used method for authentication and identification, with applications for secure access and locating missing persons. Its success is largely attributed to deep learning, which leverages large datasets and effective loss functions to learn discriminative features. Despite these advances, facial recognition still faces challenges in explainability, demographic bias, privacy, and robustness to aging, pose variations, lighting changes, occlusions, and facial expressions. Privacy regulations have also led to the degradation of several datasets, raising legal, ethical, and privacy concerns. Synthetic facial data generation has been proposed as a promising solution. It mitigates privacy issues, enables experimentation with controlled facial attributes, alleviates demographic bias, and provides supplementary data to improve models trained on real data. This study compares the effectiveness of synthetic facial datasets generated using different techniques in facial recognition tasks. We evaluate accuracy, rank-1, rank-5, and the true positive rate at a false positive rate of 0.01% on eight leading datasets, offering a comparative analysis not extensively explored in the literature. Results demonstrate the ability of synthetic data to capture realistic variations while emphasizing the need for further research to close the performance gap with real data. Techniques such as diffusion models, GANs, and 3D models show substantial progress; however, challenges remain. --- ### World Models That Know When They Don't Know - Controllable Video Generation with Calibrated Uncertainty Authors: Zhiting Mei, Tenny Yin, Micah Baker, Ola Shorinwa, Anirudha Majumdar Published: 2025-12-05 Categories: cs.CV, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2512.05927v2 Recent advances in generative video models have led to significant breakthroughs in high-fidelity video synthesis, specifically in controllable video generation where the generated video is conditioned on text and action inputs, e.g., in instruction-guided video editing and world modeling in robotics. Despite these exceptional capabilities, controllable video models often hallucinate - generating future video frames that are misaligned with physical reality - which raises serious concerns in many tasks such as robot policy evaluation and planning. However, state-of-the-art video models lack the ability to assess and express their confidence, impeding hallucination mitigation. To rigorously address this challenge, we propose C3, an uncertainty quantification (UQ) method for training continuous-scale calibrated controllable video models for dense confidence estimation at the subpatch level, precisely localizing the uncertainty in each generated video frame. Our UQ method introduces three core innovations to empower video models to estimate their uncertainty. First, our method develops a novel framework that trains video models for correctness and calibration via strictly proper scoring rules. Second, we estimate the video model's uncertainty in latent space, avoiding training instability and prohibitive training costs associated with pixel-space approaches. Third, we map the dense latent-space uncertainty to interpretable pixel-level uncertainty in the RGB space for intuitive visualization, providing high-resolution uncertainty heatmaps that identify untrustworthy regions. Through extensive experiments on large-scale robot learning datasets (Bridge and DROID) and real-world evaluations, we demonstrate that our method not only provides calibrated uncertainty estimates within the training distribution, but also enables effective out-of-distribution detection. --- ### World Models That Know When They Don't Know: Controllable Video Generation with Calibrated Uncertainty Authors: Zhiting Mei, Tenny Yin, Micah Baker, Ola Shorinwa, Anirudha Majumdar Published: 2025-12-05 Categories: cs.CV, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2512.05927v1 Recent advances in generative video models have led to significant breakthroughs in high-fidelity video synthesis, specifically in controllable video generation where the generated video is conditioned on text and action inputs, e.g., in instruction-guided video editing and world modeling in robotics. Despite these exceptional capabilities, controllable video models often hallucinate - generating future video frames that are misaligned with physical reality - which raises serious concerns in many tasks such as robot policy evaluation and planning. However, state-of-the-art video models lack the ability to assess and express their confidence, impeding hallucination mitigation. To rigorously address this challenge, we propose C3, an uncertainty quantification (UQ) method for training continuous-scale calibrated controllable video models for dense confidence estimation at the subpatch level, precisely localizing the uncertainty in each generated video frame. Our UQ method introduces three core innovations to empower video models to estimate their uncertainty. First, our method develops a novel framework that trains video models for correctness and calibration via strictly proper scoring rules. Second, we estimate the video model's uncertainty in latent space, avoiding training instability and prohibitive training costs associated with pixel-space approaches. Third, we map the dense latent-space uncertainty to interpretable pixel-level uncertainty in the RGB space for intuitive visualization, providing high-resolution uncertainty heatmaps that identify untrustworthy regions. Through extensive experiments on large-scale robot learning datasets (Bridge and DROID) and real-world evaluations, we demonstrate that our method not only provides calibrated uncertainty estimates within the training distribution, but also enables effective out-of-distribution detection. --- ### BalLOT: Balanced $k$-means clustering with optimal transport Authors: Wenyan Luo, Dustin G. Mixon Published: 2025-12-05 Categories: stat.ML, cs.DS, cs.IT, cs.LG, math.OC arXiv: https://arxiv.org/abs/2512.05926v2 We consider the fundamental problem of balanced $k$-means clustering. In particular, we introduce an optimal transport approach to alternating minimization called BalLOT, and we show that it delivers a fast and effective solution to this problem. We establish this with several theoretical guarantees and a variety of numerical experiments. On the theory front, we first prove that for generic data, BalLOT produces integral couplings at each step. Next, we perform a landscape analysis to provide theoretical guarantees for both exact and partial recoveries of planted clusters under the stochastic ball model. We also propose initialization schemes that achieve one-step recovery of planted clusters. To conclude, we present numerical experiments that corroborate our theoretical results. --- ### BalLOT: Balanced $k$-means clustering with optimal transport Authors: Wenyan Luo, Dustin G. Mixon Published: 2025-12-05 Categories: stat.ML, cs.DS, cs.IT, cs.LG, math.OC arXiv: https://arxiv.org/abs/2512.05926v1 We consider the fundamental problem of balanced $k$-means clustering. In particular, we introduce an optimal transport approach to alternating minimization called BalLOT, and we show that it delivers a fast and effective solution to this problem. We establish this with a variety of numerical experiments before proving several theoretical guarantees. First, we prove that for generic data, BalLOT produces integral couplings at each step. Next, we perform a landscape analysis to provide theoretical guarantees for both exact and partial recoveries of planted clusters under the stochastic ball model. Finally, we propose initialization schemes that achieve one-step recovery of planted clusters. --- ### To Err Is Human: Systematic Quantification of Errors in Published AI Papers via LLM Analysis Authors: Federico Bianchi, Yongchan Kwon, Zachary Izzo, Linjun Zhang, James Zou Published: 2025-12-05 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2512.05925v1 How many mistakes do published AI papers contain? Peer-reviewed publications form the foundation upon which new research and knowledge are built. Errors that persist in the literature can propagate unnoticed, creating confusion in follow-up studies and complicating reproducibility. The accelerating pace of research and the increasing demands on the peer-review system make such mistakes harder to detect and avoid. To address this, we developed a Paper Correctness Checker based on GPT-5 to systematically identify mistakes in papers previously published at top AI conferences and journals. Our analysis focuses on objective mistakes-e.g., errors in formulas, derivations, calculations, figures, and tables-that have a clearly verifiable ground truth. We intentionally exclude subjective considerations such as novelty, importance, or writing quality. We find that published papers contain a non-negligible number of objective mistakes and that the average number of mistakes per paper has increased over time-from 3.8 in NeurIPS 2021 to 5.9 in NeurIPS 2025 (55.3% increase); from 4.1 in ICLR 2018 to 5.2 in ICLR 2025; and from 5.0 in TMLR 2022/23 to 5.5 in TMLR 2025. Human experts reviewed 316 potential mistakes identified by the AI Checker and confirmed that 263 were actual mistakes, corresponding to a precision of 83.2%. While most identified issues are relatively minor, correcting them would reduce confusion in the literature and strengthen reproducibility. The AI Checker also surfaced potentially more substantive mistakes that could affect the interpretation of results. Moreover, we show that the AI Checker can propose correct fixes for 75.8% of the identified mistakes. Overall, this study highlights the potential of frontier LLMs to detect and correct objective mistakes in published papers, helping to establish a firmer foundation of knowledge. --- ### LPD: Learnable Prototypes with Diversity Regularization for Weakly Supervised Histopathology Segmentation Authors: Khang Le, Anh Mai Vu, Thi Kim Trang Vo, Ha Thach, Ngoc Bui Lam Quang, Thanh-Huy Nguyen, Minh H. N. Le, Zhu Han, Chandra Mohan, Hien Van Nguyen Published: 2025-12-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.05922v1 Weakly supervised semantic segmentation (WSSS) in histopathology reduces pixel-level labeling by learning from image-level labels, but it is hindered by inter-class homogeneity, intra-class heterogeneity, and CAM-induced region shrinkage (global pooling-based class activation maps whose activations highlight only the most distinctive areas and miss nearby class regions). Recent works address these challenges by constructing a clustering prototype bank and then refining masks in a separate stage; however, such two-stage pipelines are costly, sensitive to hyperparameters, and decouple prototype discovery from segmentation learning, limiting their effectiveness and efficiency. We propose a cluster-free, one-stage learnable-prototype framework with diversity regularization to enhance morphological intra-class heterogeneity coverage. Our approach achieves state-of-the-art (SOTA) performance on BCSS-WSSS, outperforming prior methods in mIoU and mDice. Qualitative segmentation maps show sharper boundaries and fewer mislabels, and activation heatmaps further reveal that, compared with clustering-based prototypes, our learnable prototypes cover more diverse and complementary regions within each class, providing consistent qualitative evidence for their effectiveness. --- ### NICE: Neural Implicit Craniofacial Model for Orthognathic Surgery Prediction Authors: Jiawen Yang, Yihui Cao, Xuanyu Tian, Yuyao Zhang, Hongjiang Wei Published: 2025-12-05 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2512.05920v1 Orthognathic surgery is a crucial intervention for correcting dentofacial skeletal deformities to enhance occlusal functionality and facial aesthetics. Accurate postoperative facial appearance prediction remains challenging due to the complex nonlinear interactions between skeletal movements and facial soft tissue. Existing biomechanical, parametric models and deep-learning approaches either lack computational efficiency or fail to fully capture these intricate interactions. To address these limitations, we propose Neural Implicit Craniofacial Model (NICE) which employs implicit neural representations for accurate anatomical reconstruction and surgical outcome prediction. NICE comprises a shape module, which employs region-specific implicit Signed Distance Function (SDF) decoders to reconstruct the facial surface, maxilla, and mandible, and a surgery module, which employs region-specific deformation decoders. These deformation decoders are driven by a shared surgical latent code to effectively model the complex, nonlinear biomechanical response of the facial surface to skeletal movements, incorporating anatomical prior knowledge. The deformation decoders output point-wise displacement fields, enabling precise modeling of surgical outcomes. Extensive experiments demonstrate that NICE outperforms current state-of-the-art methods, notably improving prediction accuracy in critical facial regions such as lips and chin, while robustly preserving anatomical integrity. This work provides a clinically viable tool for enhanced surgical planning and patient consultation in orthognathic procedures. --- ### A Residual Variance Matching Recursive Least Squares Filter for Real-time UAV Terrain Following Authors: Xiaobo Wu, Youmin Zhang Published: 2025-12-05 Categories: eess.SP, cs.RO, stat.ML arXiv: https://arxiv.org/abs/2512.05918v1 Accurate real-time waypoints estimation for the UAV-based online Terrain Following during wildfire patrol missions is critical to ensuring flight safety and enabling wildfire detection. However, existing real-time filtering algorithms struggle to maintain accurate waypoints under measurement noise in nonlinear and time-varying systems, posing risks of flight instability and missed wildfire detections during UAV-based terrain following. To address this issue, a Residual Variance Matching Recursive Least Squares (RVM-RLS) filter, guided by a Residual Variance Matching Estimation (RVME) criterion, is proposed to adaptively estimate the real-time waypoints of nonlinear, time-varying UAV-based terrain following systems. The proposed method is validated using a UAV-based online terrain following system within a simulated terrain environment. Experimental results show that the RVM-RLS filter improves waypoints estimation accuracy by approximately 88$\%$ compared with benchmark algorithms across multiple evaluation metrics. These findings demonstrate both the methodological advances in real-time filtering and the practical potential of the RVM-RLS filter for UAV-based online wildfire patrol. --- ### KQ-SVD: Compressing the KV Cache with Provable Guarantees on Attention Fidelity Authors: Damien Lesens, Beheshteh T. Rakhshan, Guillaume Rabusseau Published: 2025-12-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.05916v1 The Key-Value (KV) cache is central to the efficiency of transformer-based large language models (LLMs), storing previously computed vectors to accelerate inference. Yet, as sequence length and batch size grow, the cache becomes a major memory bottleneck. Prior compression methods typically apply low-rank decomposition to keys alone or attempt to jointly embed queries and keys, but both approaches neglect that attention fundamentally depends on their inner products. In this work, we prove that such strategies are suboptimal for approximating the attention matrix. We introduce KQ-SVD, a simple and computationally efficient method that directly performs an optimal low-rank decomposition of the attention matrix via a closed-form solution. By targeting the true source of redundancy, KQ-SVD preserves attention outputs with higher fidelity under compression. Extensive evaluations on LLaMA and Mistral models demonstrate that our approach consistently delivers superior projection quality. --- ### LDLT $\mathcal{L}$-Lipschitz Network: Generalized Deep End-To-End Lipschitz Network Construction Authors: Marius F. R. Juston, Ramavarapu S. Sreenivas, Dustin Nottage, Ahmet Soylemezoglu Published: 2025-12-05 Categories: cs.LG, eess.SY arXiv: https://arxiv.org/abs/2512.05915v1 Deep residual networks (ResNets) have demonstrated outstanding success in computer vision tasks, attributed to their ability to maintain gradient flow through deep architectures. Simultaneously, controlling the Lipschitz constant in neural networks has emerged as an essential area of research to enhance adversarial robustness and network certifiability. This paper presents a rigorous approach to the general design of $\mathcal{L}$-Lipschitz deep residual networks using a Linear Matrix Inequality (LMI) framework. Initially, the ResNet architecture was reformulated as a cyclic tridiagonal LMI, and closed-form constraints on network parameters were derived to ensure $\mathcal{L}$-Lipschitz continuity; however, using a new $LDL^\top$ decomposition approach for certifying LMI feasibility, we extend the construction of $\mathcal{L}$-Lipchitz networks to any other nonlinear architecture. Our contributions include a provable parameterization methodology for constructing Lipschitz-constrained residual networks and other hierarchical architectures. Cholesky decomposition is also used for efficient parameterization. These findings enable robust network designs applicable to adversarial robustness, certified training, and control systems. The $LDL^\top$ formulation is shown to be a tight relaxation of the SDP-based network, maintaining full expressiveness and achieving 3\%-13\% accuracy gains over SLL Layers on 121 UCI data sets. --- ### Natural Language Summarization Enables Multi-Repository Bug Localization by LLMs in Microservice Architectures Authors: Amirkia Rafiei Oskooei, S. Selcan Yukcu, Mehmet Cevheri Bozoglan, Mehmet S. Aktas Published: 2025-12-05 Categories: cs.SE, cs.AI, cs.CL, cs.IR arXiv: https://arxiv.org/abs/2512.05908v1 Bug localization in multi-repository microservice architectures is challenging due to the semantic gap between natural language bug reports and code, LLM context limitations, and the need to first identify the correct repository. We propose reframing this as a natural language reasoning task by transforming codebases into hierarchical NL summaries and performing NL-to-NL search instead of cross-modal retrieval. Our approach builds context-aware summaries at file, directory, and repository levels, then uses a two-phase search: first routing bug reports to relevant repositories, then performing top-down localization within those repositories. Evaluated on DNext, an industrial system with 46 repositories and 1.1M lines of code, our method achieves Pass@10 of 0.82 and MRR of 0.50, significantly outperforming retrieval baselines and agentic RAG systems like GitHub Copilot and Cursor. This work demonstrates that engineered natural language representations can be more effective than raw source code for scalable bug localization, providing an interpretable repository -> directory -> file search path, which is vital for building trust in enterprise AI tools by providing essential transparency. --- ### ADSEQ: A delay-aware autograd-compatible framework for spike-event delivery in SNNs Authors: Lennart P. L. Landsmeer, Amirreza Movahedin, Said Hamdioui, Christos Strydis Published: 2025-12-05 Categories: cs.NE arXiv: https://arxiv.org/abs/2512.05906v2 Spiking neural networks (SNNs), central to computational neuroscience and neuromorphic machine learning (ML), require efficient simulation and gradient-based training. While AI accelerators offer promising speedups, gradient-based SNNs typically implement sparse spike events using dense, memory-heavy data-structures. Existing exact gradient methods lack generality, and current simulators often omit or inefficiently handle delayed spikes. We address this by deriving gradient computation through spike event queues, including delays, and implementing this into memory-efficient, autodifferentiable spike event queues (ADSEQ). These are benchmarked across CPU, GPU, TPU, and LPU platforms. We find that queue design strongly shapes performance. CPUs, as expected, perform well with traditional tree-based or FIFO implementations, while GPUs excel with ring buffers for smaller simulations, yet under higher memory pressure prefer more sparse data-structures. TPUs seem to favor an implementation based on sorting intrinsics. Selective spike dropping provides a simple performance-accuracy trade-off, which could be enhanced by future autograd frameworks adapting diverging primal/tangent data-structures. --- ### EventQueues: Autodifferentiable spike event queues for brain simulation on AI accelerators Authors: Lennart P. L. Landsmeer, Amirreza Movahedin, Said Hamdioui, Christos Strydis Published: 2025-12-05 Categories: cs.NE arXiv: https://arxiv.org/abs/2512.05906v1 Spiking neural networks (SNNs), central to computational neuroscience and neuromorphic machine learning (ML), require efficient simulation and gradient-based training. While AI accelerators offer promising speedups, gradient-based SNNs typically implement sparse spike events using dense, memory-heavy data-structures. Existing exact gradient methods lack generality, and current simulators often omit or inefficiently handle delayed spikes. We address this by deriving gradient computation through spike event queues, including delays, and implementing memory-efficient, gradient-enabled event queue structures. These are benchmarked across CPU, GPU, TPU, and LPU platforms. We find that queue design strongly shapes performance. CPUs, as expected, perform well with traditional tree-based or FIFO implementations, while GPUs excel with ring buffers for smaller simulations, yet under higher memory pressure prefer more sparse data-structures. TPUs seem to favor an implementation based on sorting intrinsics. Selective spike dropping provides a simple performance-accuracy trade-off, which could be enhanced by future autograd frameworks adapting diverging primal/tangent data-structures. --- ### SCAIL: Towards Studio-Grade Character Animation via In-Context Learning of 3D-Consistent Pose Representations Authors: Wenhao Yan, Sheng Ye, Zhuoyi Yang, Jiayan Teng, ZhenHui Dong, Kairui Wen, Xiaotao Gu, Yong-Jin Liu, Jie Tang Published: 2025-12-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.05905v3 Achieving controllable character animation that meets studio-grade standards remains challenging despite recent progress. Existing approaches can transfer motion from a driving video to a reference image, but often fail to preserve structural fidelity and temporal consistency in wild scenarios involving complex motion and cross-identity animations. In this work, we present \textbf{SCAIL} (a framework toward \textbf{S}tudio-grade \textbf{C}haracter \textbf{A}nimation via \textbf{I}n-context \textbf{L}earning), which is designed to address these challenges from two key innovations. First, we propose a novel 3D pose representation, providing a robust and flexible motion signal. Second, we introduce a full-context pose injection mechanism within a diffusion-transformer, enabling effective spatio-temporal reasoning over full motion sequences. To align with studio-grade requirements, we develop a curated data pipeline ensuring both diversity and quality, and establish a comprehensive benchmark for systematic evaluation. Experiments show that \textbf{SCAIL} achieves state-of-the-art performance and advances character animation toward studio-grade controlling. Code and model are available at \href{https://github.com/zai-org/SCAIL}{zai-org/SCAIL}. --- ### SCAIL: Towards Studio-Grade Character Animation via In-Context Learning of 3D-Consistent Pose Representations Authors: Wenhao Yan, Sheng Ye, Zhuoyi Yang, Jiayan Teng, ZhenHui Dong, Kairui Wen, Xiaotao Gu, Yong-Jin Liu, Jie Tang Published: 2025-12-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.05905v1 Achieving character animation that meets studio-grade production standards remains challenging despite recent progress. Existing approaches can transfer motion from a driving video to a reference image, but often fail to preserve structural fidelity and temporal consistency in wild scenarios involving complex motion and cross-identity animations. In this work, we present \textbf{SCAIL} (\textbf{S}tudio-grade \textbf{C}haracter \textbf{A}nimation via \textbf{I}n-context \textbf{L}earning), a framework designed to address these challenges from two key innovations. First, we propose a novel 3D pose representation, providing a more robust and flexible motion signal. Second, we introduce a full-context pose injection mechanism within a diffusion-transformer architecture, enabling effective spatio-temporal reasoning over full motion sequences. To align with studio-level requirements, we develop a curated data pipeline ensuring both diversity and quality, and establish a comprehensive benchmark for systematic evaluation. Experiments show that \textbf{SCAIL} achieves state-of-the-art performance and advances character animation toward studio-grade reliability and realism. --- ### NeuroMemFPP: A recurrent neural approach for memory-aware parameter estimation in fractional Poisson process Authors: Neha Gupta, Aditya Maheshwari Published: 2025-12-05 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2512.05893v1 In this paper, we propose a recurrent neural network (RNN)-based framework for estimating the parameters of the fractional Poisson process (FPP), which models event arrivals with memory and long-range dependence. The Long Short-Term Memory (LSTM) network estimates the key parameters $μ>0$ and $β\in(0,1)$ from sequences of inter-arrival times, effectively modeling their temporal dependencies. Our experiments on synthetic data show that the proposed approach reduces the mean squared error (MSE) by about 55.3\% compared to the traditional method of moments (MOM) and performs reliably across different training conditions. We tested the method on two real-world high-frequency datasets: emergency call records from Montgomery County, PA, and AAPL stock trading data. The results show that the LSTM can effectively track daily patterns and parameter changes, indicating its effectiveness on real-world data with complex time dependencies. --- ### Bootstrapping Fuzzers for Compilers of Low-Resource Language Dialects Using Language Models Authors: Sairam Vaidya, Marcel Böhme, Loris D'Antoni Published: 2025-12-05 Categories: cs.SE, cs.LG, cs.PL arXiv: https://arxiv.org/abs/2512.05887v1 Modern extensible compiler frameworks-such as MLIR-enable rapid creation of domain-specific language dialects. This flexibility, however, makes correctness harder to ensure as the same extensibility that accelerates development also complicates maintaining the testing infrastructure. Extensible languages require automated test generation that is both dialect-agnostic (works across dialects without manual adaptation) and dialect-effective (targets dialect-specific features to find bugs). Existing approaches typically sacrifice one of these goals by either requiring manually constructed seed corpora for each dialect, or by failing to be effective. We present a dialect-agnostic and dialect-effective grammar-based and coverage-guided fuzzing approach for extensible compilers that combines two key insights from existing work: (i) the grammars of dialects, which already encode the structural and type constraints, can often be extracted automatically from the dialect specification; and (ii) these grammars can be used in combination with pre-trained large language models to automatically generate representative and diverse seed inputs from the full dialect space without requiring any manual input or training data. These seeds can then be used to bootstrap coverage-guided fuzzers. We built this approach into a tool, Germinator. When evaluated on six MLIR projects spanning 91 dialects, Germinator generated seeds improve line coverage by 10-120% over grammar-based baselines. We compare against grammar-based baselines because they are the only class of existing automatic seed generators that can be applied uniformly across MLIR's heterogeneous dialect ecosystem. Germinator discovers 88 previously unknown bugs (40 confirmed), including 23 in dialects with no prior automated test generators, demonstrating effective and controllable testing of low-resource dialects at scale. --- ### DAE-HardNet: A Physics Constrained Neural Network Enforcing Differential-Algebraic Hard Constraints Authors: Rahul Golder, Bimol Nath Roy, M. M. Faruque Hasan Published: 2025-12-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.05881v1 Traditional physics-informed neural networks (PINNs) do not always satisfy physics based constraints, especially when the constraints include differential operators. Rather, they minimize the constraint violations in a soft way. Strict satisfaction of differential-algebraic equations (DAEs) to embed domain knowledge and first-principles in data-driven models is generally challenging. This is because data-driven models consider the original functions to be black-box whose derivatives can only be obtained after evaluating the functions. We introduce DAE-HardNet, a physics-constrained (rather than simply physics-informed) neural network that learns both the functions and their derivatives simultaneously, while enforcing algebraic as well as differential constraints. This is done by projecting model predictions onto the constraint manifold using a differentiable projection layer. We apply DAE-HardNet to several systems and test problems governed by DAEs, including the dynamic Lotka-Volterra predator-prey system and transient heat conduction. We also show the ability of DAE-HardNet to estimate unknown parameters through a parameter estimation problem. Compared to multilayer perceptrons (MLPs) and PINNs, DAE-HardNet achieves orders of magnitude reduction in the physics loss while maintaining the prediction accuracy. It has the added benefits of learning the derivatives which improves the constrained learning of the backbone neural network prior to the projection layer. For specific problems, this suggests that the projection layer can be bypassed for faster inference. The current implementation and codes are available at https://github.com/SOULS-TAMU/DAE-HardNet. --- ### Neural Coherence : Find higher performance to out-of-distribution tasks from few samples Authors: Simon Guiroy, Mats Richter, Sarath Chandar, Christopher Pal Published: 2025-12-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.05880v1 To create state-of-the-art models for many downstream tasks, it has become common practice to fine-tune a pre-trained large vision model. However, it remains an open question of how to best determine which of the many possible model checkpoints resulting from a large training run to use as the starting point. This becomes especially important when data for the target task of interest is scarce, unlabeled and out-of-distribution. In such scenarios, common methods relying on in-distribution validation data become unreliable or inapplicable. This work proposes a novel approach for model selection that operates reliably on just a few unlabeled examples from the target task. Our approach is based on a novel concept: Neural Coherence, which entails characterizing a model's activation statistics for source and target domains, allowing one to define model selection methods with high data-efficiency. We provide experiments where models are pre-trained on ImageNet1K and examine target domains consisting of Food-101, PlantNet-300K and iNaturalist. We also evaluate it in many meta-learning settings. Our approach significantly improves generalization across these different target domains compared to established baselines. We further demonstrate the versatility of Neural Coherence as a powerful principle by showing its effectiveness in training data selection. --- ### Computational Design of Low-Volatility Lubricants for Space Using Interpretable Machine Learning Authors: Daniel Miliate, Ashlie Martini Published: 2025-12-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.05870v1 The function and lifetime of moving mechanical assemblies (MMAs) in space depend on the properties of lubricants. MMAs that experience high speeds or high cycles require liquid based lubricants due to their ability to reflow to the point of contact. However, only a few liquid-based lubricants have vapor pressures low enough for the vacuum conditions of space, each of which has limitations that add constraints to MMA designs. This work introduces a data-driven machine learning (ML) approach to predicting vapor pressure, enabling virtual screening and discovery of new space-suitable liquid lubricants. The ML models are trained with data from both high-throughput molecular dynamics simulations and experimental databases. The models are designed to prioritize interpretability, enabling the relationships between chemical structure and vapor pressure to be identified. Based on these insights, several candidate molecules are proposed that may have promise for future space lubricant applications in MMAs. --- ### Predicting Price Movements in High-Frequency Financial Data with Spiking Neural Networks Authors: Brian Ezinwoke, Oliver Rhodes Published: 2025-12-05 Categories: cs.LG, q-fin.CP arXiv: https://arxiv.org/abs/2512.05868v1 Modern high-frequency trading (HFT) environments are characterized by sudden price spikes that present both risk and opportunity, but conventional financial models often fail to capture the required fine temporal structure. Spiking Neural Networks (SNNs) offer a biologically inspired framework well-suited to these challenges due to their natural ability to process discrete events and preserve millisecond-scale timing. This work investigates the application of SNNs to high-frequency price-spike forecasting, enhancing performance via robust hyperparameter tuning with Bayesian Optimization (BO). This work converts high-frequency stock data into spike trains and evaluates three architectures: an established unsupervised STDP-trained SNN, a novel SNN with explicit inhibitory competition, and a supervised backpropagation network. BO was driven by a novel objective, Penalized Spike Accuracy (PSA), designed to ensure a network's predicted price spike rate aligns with the empirical rate of price events. Simulated trading demonstrated that models optimized with PSA consistently outperformed their Spike Accuracy (SA)-tuned counterparts and baselines. Specifically, the extended SNN model with PSA achieved the highest cumulative return (76.8%) in simple backtesting, significantly surpassing the supervised alternative (42.54% return). These results validate the potential of spiking networks, when robustly tuned with task-specific objectives, for effective price spike forecasting in HFT. --- ### Underwater Image Reconstruction Using a Swin Transformer-Based Generator and PatchGAN Discriminator Authors: Md. Mahbub Hasan Akash, Aria Tasnim Mridula, Sheekar Banerjee, Ishtiak Al Mamoon Published: 2025-12-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.05866v1 Underwater imaging is essential for marine exploration, environmental monitoring, and infrastructure inspection. However, water causes severe image degradation through wavelength-dependent absorption and scattering, resulting in color distortion, low contrast, and haze effects. Traditional reconstruction methods and convolutional neural network-based approaches often fail to adequately address these challenges due to limited receptive fields and inability to model global dependencies. This paper presented a novel deep learning framework that integrated a Swin Transformer architecture within a generative adversarial network (GAN) for underwater image reconstruction. Our generator employed a U-Net structure with Swin Transformer blocks to capture both local features and long-range dependencies crucial for color correction across entire images. A PatchGAN discriminator provided adversarial training to ensure high-frequency detail preservation. We trained and evaluated our model on the EUVP dataset, which contains paired underwater images of varying quality. Quantitative results demonstrate stateof-the-art performance with PSNR of 24.76 dB and SSIM of 0.89, representing significant improvements over existing methods. Visual results showed effective color balance restoration, contrast improvement, and haze reduction. An ablation study confirms the superiority of our Swin Transformer designed over convolutional alternatives. The proposed method offers robust underwater image reconstruction suitable for various marine applications. --- ### Intrinsically Interpretable Attention via Sparse Post-Training Authors: Florent Draye, Anson Lei, Hsiao-Ru Pan, Ingmar Posner, Bernhard Schölkopf Published: 2025-12-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.05865v5 We introduce a simple post-training method that makes transformer attention sparse without sacrificing performance. Applying a flexible sparsity regularisation under a constrained-loss objective, we show on models up to 7B parameters that it is possible to retain the original pretraining loss while reducing attention connectivity to $\approx 0.4 \%$ of its edges. Unlike sparse-attention methods designed for computational efficiency, our approach leverages sparsity as a structural prior: it preserves capability while exposing a more organized and interpretable connectivity pattern. We find that this local sparsity cascades into global circuit simplification: task-specific circuits involve far fewer components (attention heads and MLPs) with up to 100x fewer edges connecting them. Additionally, using cross-layer transcoders, we show that sparse attention substantially simplifies attention attribution, enabling a unified view of feature-based and circuit-based perspectives. These results demonstrate that transformer attention can be made orders of magnitude sparser, suggesting that much of its computation is redundant and that sparsity may serve as a guiding principle for more structured and interpretable models. --- ### Sparse Attention Post-Training for Mechanistic Interpretability Authors: Florent Draye, Anson Lei, Ingmar Posner, Bernhard Schölkopf Published: 2025-12-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.05865v1 We introduce a simple post-training method that makes transformer attention sparse without sacrificing performance. Applying a flexible sparsity regularisation under a constrained-loss objective, we show on models up to 1B parameters that it is possible to retain the original pretraining loss while reducing attention connectivity to $\approx 0.3 \%$ of its edges. Unlike sparse-attention methods designed for computational efficiency, our approach leverages sparsity as a structural prior: it preserves capability while exposing a more organized and interpretable connectivity pattern. We find that this local sparsity cascades into global circuit simplification: task-specific circuits involve far fewer components (attention heads and MLPs) with up to 100x fewer edges connecting them. These results demonstrate that transformer attention can be made orders of magnitude sparser, suggesting that much of its computation is redundant and that sparsity may serve as a guiding principle for more structured and interpretable models. --- ### Optimizing Medical Question-Answering Systems: A Comparative Study of Fine-Tuned and Zero-Shot Large Language Models with RAG Framework Authors: Tasnimul Hassan, Md Faisal Karim, Haziq Jeelani, Elham Behnam, Robert Green, Fayeq Jeelani Syed Published: 2025-12-05 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.05863v1 Medical question-answering (QA) systems can benefit from advances in large language models (LLMs), but directly applying LLMs to the clinical domain poses challenges such as maintaining factual accuracy and avoiding hallucinations. In this paper, we present a retrieval-augmented generation (RAG) based medical QA system that combines domain-specific knowledge retrieval with open-source LLMs to answer medical questions. We fine-tune two state-of-the-art open LLMs (LLaMA~2 and Falcon) using Low-Rank Adaptation (LoRA) for efficient domain specialization. The system retrieves relevant medical literature to ground the LLM's answers, thereby improving factual correctness and reducing hallucinations. We evaluate the approach on benchmark datasets (PubMedQA and MedMCQA) and show that retrieval augmentation yields measurable improvements in answer accuracy compared to using LLMs alone. Our fine-tuned LLaMA~2 model achieves 71.8% accuracy on PubMedQA, substantially improving over the 55.4% zero-shot baseline, while maintaining transparency by providing source references. We also detail the system design and fine-tuning methodology, demonstrating that grounding answers in retrieved evidence reduces unsupported content by approximately 60%. These results highlight the potential of RAG-augmented open-source LLMs for reliable biomedical QA, pointing toward practical clinical informatics applications. --- ### Edit-aware RAW Reconstruction Authors: Abhijith Punnappurath, Luxi Zhao, Ke Zhao, Hue Nguyen, Radek Grzeszczuk, Michael S. Brown Published: 2025-12-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.05859v2 Users frequently edit camera images post-capture to achieve their preferred photofinishing style. While editing in the RAW domain provides greater accuracy and flexibility, most edits are performed on the camera's display-referred output (e.g., 8-bit sRGB JPEG) since RAW images are rarely stored. Existing RAW reconstruction methods can recover RAW data from sRGB images, but these approaches are typically optimized for pixel-wise RAW reconstruction fidelity and tend to degrade under diverse rendering styles and editing operations. We introduce a plug-and-play, edit-aware loss function that can be integrated into any existing RAW reconstruction framework to make the recovered RAWs more robust to different rendering styles and edits. Our loss formulation incorporates a modular, differentiable image signal processor (ISP) that simulates realistic photofinishing pipelines with tunable parameters. During training, parameters for each ISP module are randomly sampled from carefully designed distributions that model practical variations in real camera processing. The loss is then computed in sRGB space between ground-truth and reconstructed RAWs rendered through this differentiable ISP. Incorporating our loss improves sRGB reconstruction quality by up to 1.5-2 dB PSNR across various editing conditions. Moreover, when applied to metadata-assisted RAW reconstruction methods, our approach enables fine-tuning for target edits, yielding further gains. Since photographic editing is the primary motivation for RAW reconstruction in consumer imaging, our simple yet effective loss function provides a general mechanism for enhancing edit fidelity and rendering flexibility across existing methods. --- ### Edit-aware RAW Reconstruction Authors: Abhijith Punnappurath, Luxi Zhao, Ke Zhao, Hue Nguyen, Radek Grzeszczuk, Michael S. Brown Published: 2025-12-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.05859v1 Users frequently edit camera images post-capture to achieve their preferred photofinishing style. While editing in the RAW domain provides greater accuracy and flexibility, most edits are performed on the camera's display-referred output (e.g., 8-bit sRGB JPEG) since RAW images are rarely stored. Existing RAW reconstruction methods can recover RAW data from sRGB images, but these approaches are typically optimized for pixel-wise RAW reconstruction fidelity and tend to degrade under diverse rendering styles and editing operations. We introduce a plug-and-play, edit-aware loss function that can be integrated into any existing RAW reconstruction framework to make the recovered RAWs more robust to different rendering styles and edits. Our loss formulation incorporates a modular, differentiable image signal processor (ISP) that simulates realistic photofinishing pipelines with tunable parameters. During training, parameters for each ISP module are randomly sampled from carefully designed distributions that model practical variations in real camera processing. The loss is then computed in sRGB space between ground-truth and reconstructed RAWs rendered through this differentiable ISP. Incorporating our loss improves sRGB reconstruction quality by up to 1.5-2 dB PSNR across various editing conditions. Moreover, when applied to metadata-assisted RAW reconstruction methods, our approach enables fine-tuning for target edits, yielding further gains. Since photographic editing is the primary motivation for RAW reconstruction in consumer imaging, our simple yet effective loss function provides a general mechanism for enhancing edit fidelity and rendering flexibility across existing methods. --- ### Prompting Science Report 4: Playing Pretend: Expert Personas Don't Improve Factual Accuracy Authors: Savir Basil, Ina Shapiro, Dan Shapiro, Ethan Mollick, Lilach Mollick, Lennart Meincke Published: 2025-12-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.05858v1 This is the fourth in a series of short reports that help business, education, and policy leaders understand the technical details of working with AI through rigorous testing. Here, we ask whether assigning personas to models improves performance on difficult objective multiple-choice questions. We study both domain-specific expert personas and low-knowledge personas, evaluating six models on GPQA Diamond (Rein et al. 2024) and MMLU-Pro (Wang et al. 2024), graduate-level questions spanning science, engineering, and law. We tested three approaches: -In-Domain Experts: Assigning the model an expert persona ("you are a physics expert") matched to the problem type (physics problems) had no significant impact on performance (with the exception of the Gemini 2.0 Flash model). -Off-Domain Experts (Domain-Mismatched): Assigning the model an expert persona ("you are a physics expert") not matched to the problem type (law problems) resulted in marginal differences. -Low-Knowledge Personas: We assigned the model negative capability personas (layperson, young child, toddler), which were generally harmful to benchmark accuracy. Across both benchmarks, persona prompts generally did not improve accuracy relative to a no-persona baseline. Expert personas showed no consistent benefit across models, with few exceptions. Domain-mismatched expert personas sometimes degraded performance. Low-knowledge personas often reduced accuracy. These results are about the accuracy of answers only; personas may serve other purposes (such as altering the tone of outputs), beyond improving factual performance. --- ### VRSA: Jailbreaking Multimodal Large Language Models through Visual Reasoning Sequential Attack Authors: Shiji Zhao, Shukun Xiong, Yao Huang, Yan Jin, Zhenyu Wu, Jiyang Guan, Ranjie Duan, Jialing Tao, Hui Xue, Xingxing Wei Published: 2025-12-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.05853v2 Multimodal Large Language Models (MLLMs) are widely used in various fields due to their powerful cross-modal comprehension and generation capabilities. However, more modalities bring more vulnerabilities to being utilized for jailbreak attacks, which induces MLLMs to output harmful content. Due to the strong reasoning ability of MLLMs, previous jailbreak attacks try to explore reasoning safety risk in text modal, while similar threats have been largely overlooked in the visual modal. To fully evaluate potential safety risks in the visual reasoning task, we propose Visual Reasoning Sequential Attack (VRSA), which induces MLLMs to gradually externalize and aggregate complete harmful intent by decomposing the original harmful text into several sequentially related sub-images. In particular, to enhance the rationality of the scene in the image sequence, we propose Adaptive Scene Refinement to optimize the scene most relevant to the original harmful query. To ensure the semantic continuity of the generated image, we propose Semantic Coherent Completion to iteratively rewrite each sub-text combined with contextual information in this scene. In addition, we propose Text-Image Consistency Alignment to keep the semantical consistency. A series of experiments demonstrates that the VRSA can achieve a higher attack success rate compared with the state-of-the-art jailbreak attack methods on both the open-source and closed-source MLLMs such as GPT-4o and Claude-4.5-Sonnet. --- ### VRSA: Jailbreaking Multimodal Large Language Models through Visual Reasoning Sequential Attack Authors: Shiji Zhao, Shukun Xiong, Yao Huang, Yan Jin, Zhenyu Wu, Jiyang Guan, Ranjie Duan, Jialing Tao, Hui Xue, Xingxing Wei Published: 2025-12-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.05853v1 Multimodal Large Language Models (MLLMs) are widely used in various fields due to their powerful cross-modal comprehension and generation capabilities. However, more modalities bring more vulnerabilities to being utilized for jailbreak attacks, which induces MLLMs to output harmful content. Due to the strong reasoning ability of MLLMs, previous jailbreak attacks try to explore reasoning safety risk in text modal, while similar threats have been largely overlooked in the visual modal. To fully evaluate potential safety risks in the visual reasoning task, we propose Visual Reasoning Sequential Attack (VRSA), which induces MLLMs to gradually externalize and aggregate complete harmful intent by decomposing the original harmful text into several sequentially related sub-images. In particular, to enhance the rationality of the scene in the image sequence, we propose Adaptive Scene Refinement to optimize the scene most relevant to the original harmful query. To ensure the semantic continuity of the generated image, we propose Semantic Coherent Completion to iteratively rewrite each sub-text combined with contextual information in this scene. In addition, we propose Text-Image Consistency Alignment to keep the semantical consistency. A series of experiments demonstrates that the VRSA can achieve a higher attack success rate compared with the state-of-the-art jailbreak attack methods on both the open-source and closed-source MLLMs such as GPT-4o and Claude-4.5-Sonnet. ---