Authors: Mahdi Samiei, Mahdi Mansouri, Mahdieh Soleymani Baghshah
Abstract: Large language models (LLMs) have achieved remarkable results on tasks framed as reasoning problems, yet their true ability to perform procedural reasoning, executing multi-step, rule-based computations remains unclear. Unlike algorithmic systems, which can deterministically execute long-horizon symbolic procedures, LLMs often degrade under extended reasoning chains, but there is no controlled, interpretable benchmark to isolate and measure this collapse. We introduce Finite-State Machine (FSM) Execution as a minimal, fully interpretable framework for evaluating the procedural reasoning capacity of LLMs. In our setup, the model is given an explicit FSM definition and must execute it step-by-step given input actions, maintaining state consistency over multiple turns. This task requires no world knowledge, only faithful application of deterministic transition rules, making it a direct probe of the model's internal procedural fidelity. We measure both Turn Accuracy and Task Accuracy to disentangle immediate computation from cumulative state maintenance. Empirical results reveal systematic degradation as task horizon or branching complexity increases. Models perform significantly worse when rule retrieval involves high branching factors than when memory span is long. Larger models show improved local accuracy but remain brittle under multi-step reasoning unless explicitly prompted to externalize intermediate steps. FSM-based evaluation offers a transparent, complexity-controlled probe for diagnosing this failure mode and guiding the design of inductive biases that enable genuine long-horizon procedural competence. By grounding reasoning in measurable execution fidelity rather than surface correctness, this work helps establish a rigorous experimental foundation for understanding and improving the algorithmic reliability of LLMs.
Authors: George Tsoukalas, Rahul Saha, Amitayush Thakur, Sabrina Reguyal, Swarat Chaudhuri
Abstract: We take two key steps in automating the open-ended discovery of new mathematical theories, a grand challenge in artificial intelligence. First, we introduce $\emph{FERMAT}$, a reinforcement learning (RL) environment that models concept discovery and theorem-proving using a set of symbolic actions, opening up a range of RL problems relevant to theory discovery. Second, we explore a specific problem through $\emph{FERMAT}$: automatically scoring the $\emph{interestingness}$ of mathematical objects. We investigate evolutionary algorithms for synthesizing nontrivial interestingness measures. In particular, we introduce an LLM-based evolutionary algorithm that features function abstraction, leading to notable improvements in discovering elementary number theory and finite fields over hard-coded baselines. We open-source the $\emph{FERMAT}$ environment at this URL(https://github.com/trishullab/Fermat).
Authors: Keith Moore, Jun W. Kim, David Lyu, Jeffrey Heo, Ehsan Adeli
Abstract: We present Ask WhAI, a systems-level framework for inspecting and perturbing belief states in multi-agent interactions. The framework records and replays agent interactions, supports out-of-band queries into each agent's beliefs and rationale, and enables counterfactual evidence injection to test how belief structures respond to new information. We apply the framework to a medical case simulator notable for its multi-agent shared memory (a time-stamped electronic medical record, or EMR) and an oracle agent (the LabAgent) that holds ground truth lab results revealed only when explicitly queried. We stress-test the system on a multi-specialty diagnostic journey for a child with an abrupt-onset neuropsychiatric presentation. Large language model agents, each primed with strong role-specific priors ("act like a neurologist", "act like an infectious disease specialist"), write to a shared medical record and interact with a moderator across sequential or parallel encounters. Breakpoints at key diagnostic moments enable pre- and post-event belief queries, allowing us to distinguish entrenched priors from reasoning or evidence-integration effects. The simulation reveals that agent beliefs often mirror real-world disciplinary stances, including overreliance on canonical studies and resistance to counterevidence, and that these beliefs can be traced and interrogated in ways not possible with human experts. By making such dynamics visible and testable, Ask WhAI offers a reproducible way to study belief formation and epistemic silos in multi-agent scientific reasoning.
Authors: Michele Ronco, Damien Delforge, Wiebke S. J\"ager, Christina Corbane
Abstract: Subnational location data of disaster events are critical for risk assessment and disaster risk reduction. Disaster databases such as EM-DAT often report locations in unstructured textual form, with inconsistent granularity or spelling, that make it difficult to integrate with spatial datasets. We present a fully automated LLM-assisted workflow that processes and cleans textual location information using GPT-4o, and assigns geometries by cross-checking three independent geoinformation repositories: GADM, OpenStreetMap and Wikidata. Based on the agreement and availability of these sources, we assign a reliability score to each location while generating subnational geometries. Applied to the EM-DAT dataset from 2000 to 2024, the workflow geocodes 14,215 events across 17,948 unique locations. Unlike previous methods, our approach requires no manual intervention, covers all disaster types, enables cross-verification across multiple sources, and allows flexible remapping to preferred frameworks. Beyond the dataset, we demonstrate the potential of LLMs to extract and structure geographic information from unstructured text, offering a scalable and reliable method for related analyses.
Authors: Martin Monperrus, Benoit Baudry, Cl\'ement Vidal
Abstract: This paper documents Project Rachel, an action research study that created and tracked a complete AI academic identity named Rachel So. Through careful publication of AI-generated research papers, we investigate how the scholarly ecosystem responds to AI authorship. Rachel So published 10+ papers between March and October 2025, was cited, and received a peer review invitation. We discuss the implications of AI authorship on publishers, researchers, and the scientific system at large. This work contributes empirical action research data to the necessary debate about the future of scholarly communication with super human, hyper capable AI systems.
Authors: Robab Aghazadeh Chakherlou, Siddartha Khastgir, Xingyu Zhao, Jerein Jeyachandran, Shufeng Chen
Abstract: Assuring the trustworthiness and safety of AI systems, e.g., autonomous vehicles (AV), depends critically on the data-related safety properties, e.g., representativeness, completeness, etc., of the datasets used for their training and testing. Among these properties, this paper focuses on representativeness-the extent to which the scenario-based data used for training and testing, reflect the operational conditions that the system is designed to operate safely in, i.e., Operational Design Domain (ODD) or expected to encounter, i.e., Target Operational Domain (TOD). We propose a probabilistic method that quantifies representativeness by comparing the statistical distribution of features encoded by the scenario suites with the corresponding distribution of features representing the TOD, acknowledging that the true TOD distribution is unknown, as it can only be inferred from limited data. We apply an imprecise Bayesian method to handle limited data and uncertain priors. The imprecise Bayesian formulation produces interval-valued, uncertainty-aware estimates of representativeness, rather than a single value. We present a numerical example comparing the distributions of the scenario suite and the inferred TOD across operational categories-weather, road type, time of day, etc., under dependencies and prior uncertainty. We estimate representativeness locally (between categories) and globally as an interval.
Authors: Fatemeh Lotfi, Hossein Rajoli, Fatemeh Afghah
Abstract: Next-generation networks utilize the Open Radio Access Network (O-RAN) architecture to enable dynamic resource management, facilitated by the RAN Intelligent Controller (RIC). While deep reinforcement learning (DRL) models show promise in optimizing network resources, they often struggle with robustness and generalizability in dynamic environments. This paper introduces a novel resource management approach that enhances the Soft Actor Critic (SAC) algorithm with Sharpness-Aware Minimization (SAM) in a distributed Multi-Agent RL (MARL) framework. Our method introduces an adaptive and selective SAM mechanism, where regularization is explicitly driven by temporal-difference (TD)-error variance, ensuring that only agents facing high environmental complexity are regularized. This targeted strategy reduces unnecessary overhead, improves training stability, and enhances generalization without sacrificing learning efficiency. We further incorporate a dynamic $\rho$ scheduling scheme to refine the exploration-exploitation trade-off across agents. Experimental results show our method significantly outperforms conventional DRL approaches, yielding up to a $22\%$ improvement in resource allocation efficiency and ensuring superior QoS satisfaction across diverse O-RAN slices.
Authors: Jian-Ting Guo, Yu-Cheng Chen, Ping-Chun Hsieh, Kuo-Hao Ho, Po-Wei Huang, Ti-Rong Wu, I-Chen Wu
Abstract: Human-like agents have long been one of the goals in pursuing artificial intelligence. Although reinforcement learning (RL) has achieved superhuman performance in many domains, relatively little attention has been focused on designing human-like RL agents. As a result, many reward-driven RL agents often exhibit unnatural behaviors compared to humans, raising concerns for both interpretability and trustworthiness. To achieve human-like behavior in RL, this paper first formulates human-likeness as trajectory optimization, where the objective is to find an action sequence that closely aligns with human behavior while also maximizing rewards, and adapts the classic receding-horizon control to human-like learning as a tractable and efficient implementation. To achieve this, we introduce Macro Action Quantization (MAQ), a human-like RL framework that distills human demonstrations into macro actions via Vector-Quantized VAE. Experiments on D4RL Adroit benchmarks show that MAQ significantly improves human-likeness, increasing trajectory similarity scores, and achieving the highest human-likeness rankings among all RL agents in the human evaluation study. Our results also demonstrate that MAQ can be easily integrated into various off-the-shelf RL algorithms, opening a promising direction for learning human-like RL agents. Our code is available at https://rlg.iis.sinica.edu.tw/papers/MAQ.
Authors: Haodong Chen, Guido Zuccon, Teerapong Leelanupab
Abstract: Genomic question answering often requires complex reasoning and integration across diverse biomedical sources. GeneGPT addressed this challenge by combining domain-specific APIs with OpenAI's code-davinci-002 large language model to enable natural language interaction with genomic databases. However, its reliance on a proprietary model limits scalability, increases operational costs, and raises concerns about data privacy and generalization. In this work, we revisit and reproduce GeneGPT in a pilot study using open source models, including Llama 3.1, Qwen2.5, and Qwen2.5 Coder, within a monolithic architecture; this allows us to identify the limitations of this approach. Building on this foundation, we then develop OpenBioLLM, a modular multi-agent framework that extends GeneGPT by introducing agent specialization for tool routing, query generation, and response validation. This enables coordinated reasoning and role-based task execution. OpenBioLLM matches or outperforms GeneGPT on over 90% of the benchmark tasks, achieving average scores of 0.849 on Gene-Turing and 0.830 on GeneHop, while using smaller open-source models without additional fine-tuning or tool-specific pretraining. OpenBioLLM's modular multi-agent design reduces latency by 40-50% across benchmark tasks, significantly improving efficiency without compromising model capability. The results of our comprehensive evaluation highlight the potential of open-source multi-agent systems for genomic question answering. Code and resources are available at https://github.com/ielab/OpenBioLLM.
Authors: Haoyong Wu, Yongmei Liu
Abstract: In this paper, we propose ProRAC (Progression-based Reasoning about Actions and Change), a neuro-symbolic framework that leverages LLMs to tackle RAC problems. ProRAC extracts fundamental RAC elements including actions and questions from the problem, progressively executes each action to derive the final state, and then evaluates the query against the progressed state to arrive at an answer. We evaluate ProRAC on several RAC benchmarks, and the results demonstrate that our approach achieves strong performance across different benchmarks, domains, LLM backbones, and types of RAC tasks.
Authors: Henrik Bradland, Morten Goodwin, Vladimir I. Zadorozhny, Per-Arne Andersen
Abstract: The performance of machine learning models on tabular data is critically dependent on high-quality feature engineering. While Large Language Models (LLMs) have shown promise in automating feature extraction (AutoFE), existing methods are often limited by monolithic LLM architectures, simplistic quantitative feedback, and a failure to systematically integrate external domain knowledge. This paper introduces Rogue One, a novel, LLM-based multi-agent framework for knowledge-informed automatic feature extraction. Rogue One operationalizes a decentralized system of three specialized agents-Scientist, Extractor, and Tester-that collaborate iteratively to discover, generate, and validate predictive features. Crucially, the framework moves beyond primitive accuracy scores by introducing a rich, qualitative feedback mechanism and a "flooding-pruning" strategy, allowing it to dynamically balance feature exploration and exploitation. By actively incorporating external knowledge via an integrated retrieval-augmented (RAG) system, Rogue One generates features that are not only statistically powerful but also semantically meaningful and interpretable. We demonstrate that Rogue One significantly outperforms state-of-the-art methods on a comprehensive suite of 19 classification and 9 regression datasets. Furthermore, we show qualitatively that the system surfaces novel, testable hypotheses, such as identifying a new potential biomarker in the myocardial dataset, underscoring its utility as a tool for scientific discovery.
Authors: Xin Gao, Shaohan Yu, Zerui Chen, Yueming Lyu, Weichen Yu, Guanghao Li, Jiyao Liu, Jianxiong Gao, Jian Liang, Ziwei Liu, Chenyang Si
Abstract: Large Reasoning Models (LRMs) improve answer quality through explicit chain-of-thought, yet this very capability introduces new safety risks: harmful content can be subtly injected, surface gradually, or be justified by misleading rationales within the reasoning trace. Existing safety evaluations, however, primarily focus on output-level judgments and rarely capture these dynamic risks along the reasoning process. In this paper, we present SafeRBench, the first benchmark that assesses LRM safety end-to-end -- from inputs and intermediate reasoning to final outputs. (1) Input Characterization: We pioneer the incorporation of risk categories and levels into input design, explicitly accounting for affected groups and severity, and thereby establish a balanced prompt suite reflecting diverse harm gradients. (2) Fine-Grained Output Analysis: We introduce a micro-thought chunking mechanism to segment long reasoning traces into semantically coherent units, enabling fine-grained evaluation across ten safety dimensions. (3) Human Safety Alignment: We validate LLM-based evaluations against human annotations specifically designed to capture safety judgments. Evaluations on 19 LRMs demonstrate that SafeRBench enables detailed, multidimensional safety assessment, offering insights into risks and protective mechanisms from multiple perspectives.
Authors: Zhiyi Duan, Zixing Shi, Hongyu Yuan, Qi Wang
Abstract: Knowledge Tracing (KT) aims to mine students' evolving knowledge states and predict their future question-answering performance. Existing methods based on heterogeneous information networks (HINs) are prone to introducing noises due to manual or random selection of meta-paths and lack necessary quality assessment of meta-path instances. Conversely, recent large language models (LLMs)-based methods ignore the rich information across students, and both paradigms struggle to deliver consistently accurate and evidence-based explanations. To address these issues, we propose an innovative framework, HIN-LLM Synergistic Enhanced Knowledge Tracing (HISE-KT), which seamlessly integrates HINs with LLMs. HISE-KT first builds a multi-relationship HIN containing diverse node types to capture the structural relations through multiple meta-paths. The LLM is then employed to intelligently score and filter meta-path instances and retain high-quality paths, pioneering automated meta-path quality assessment. Inspired by educational psychology principles, a similar student retrieval mechanism based on meta-paths is designed to provide a more valuable context for prediction. Finally, HISE-KT uses a structured prompt to integrate the target student's history with the retrieved similar trajectories, enabling the LLM to generate not only accurate predictions but also evidence-backed, explainable analysis reports. Experiments on four public datasets show that HISE-KT outperforms existing KT baselines in both prediction performance and interpretability.
Authors: Haodong Li, Jingqi Zhang, Xiao Cheng, Peihua Mai, Haoyu Wang, Yang Pan
Abstract: The remarkable language ability of Large Language Models (LLMs) stems from extensive training on vast datasets, often including copyrighted material, which raises serious concerns about unauthorized use. While Membership Inference Attacks (MIAs) offer potential solutions for detecting such violations, existing approaches face critical limitations and challenges due to LLMs' inherent overconfidence, limited access to ground truth training data, and reliance on empirically determined thresholds. We present COPYCHECK, a novel framework that leverages uncertainty signals to detect whether copyrighted content was used in LLM training sets. Our method turns LLM overconfidence from a limitation into an asset by capturing uncertainty patterns that reliably distinguish between ``seen" (training data) and ``unseen" (non-training data) content. COPYCHECK further implements a two-fold strategy: (1) strategic segmentation of files into smaller snippets to reduce dependence on large-scale training data, and (2) uncertainty-guided unsupervised clustering to eliminate the need for empirically tuned thresholds. Experiment results show that COPYCHECK achieves an average balanced accuracy of 90.1% on LLaMA 7b and 91.6% on LLaMA2 7b in detecting seen files. Compared to the SOTA baseline, COPYCHECK achieves over 90% relative improvement, reaching up to 93.8\% balanced accuracy. It further exhibits strong generalizability across architectures, maintaining high performance on GPT-J 6B. This work presents the first application of uncertainty for copyright detection in LLMs, offering practical tools for training data transparency.
Authors: Yinsheng Wang, Tario G You, L\'eonard Boussioux, Shan Liu
Abstract: This paper introduces SOLID (Synergizing Optimization and Large Language Models for Intelligent Decision-Making), a novel framework that integrates mathematical optimization with the contextual capabilities of large language models (LLMs). SOLID facilitates iterative collaboration between optimization and LLMs agents through dual prices and deviation penalties. This interaction improves the quality of the decisions while maintaining modularity and data privacy. The framework retains theoretical convergence guarantees under convexity assumptions, providing insight into the design of LLMs prompt. To evaluate SOLID, we applied it to a stock portfolio investment case with historical prices and financial news as inputs. Empirical results demonstrate convergence under various scenarios and indicate improved annualized returns compared to a baseline optimizer-only method, validating the synergy of the two agents. SOLID offers a promising framework for advancing automated and intelligent decision-making across diverse domains.
Authors: Philipp Wiesner, Daniel W. O'Neill, Francesca Larosa, Odej Kao
Abstract: AI research is increasingly moving toward complex problem solving, where models are optimized not only for pattern recognition but for multi-step reasoning. Historically, computing's global energy footprint has been stabilized by sustained efficiency gains and natural saturation thresholds in demand. But as efficiency improvements are approaching physical limits, emerging reasoning AI lacks comparable saturation points: performance is no longer limited by the amount of available training data but continues to scale with exponential compute investments in both training and inference. This paper argues that efficiency alone will not lead to sustainable reasoning AI and discusses research and policy directions to embed explicit limits into the optimization and governance of such systems.
Authors: Ninell Oldenburg, Ruchira Dhar, Anders S{\o}gaard
Abstract: In this paper, we argue that current AI research operates on a spectrum between two different underlying conceptions of intelligence: Intelligence Realism, which holds that intelligence represents a single, universal capacity measurable across all systems, and Intelligence Pluralism, which views intelligence as diverse, context-dependent capacities that cannot be reduced to a single universal measure. Through an analysis of current debates in AI research, we demonstrate how the conceptions remain largely implicit yet fundamentally shape how empirical evidence gets interpreted across a wide range of areas. These underlying views generate fundamentally different research approaches across three areas. Methodologically, they produce different approaches to model selection, benchmark design, and experimental validation. Interpretively, they lead to contradictory readings of the same empirical phenomena, from capability emergence to system limitations. Regarding AI risk, they generate categorically different assessments: realists view superintelligence as the primary risk and search for unified alignment solutions, while pluralists see diverse threats across different domains requiring context-specific solutions. We argue that making explicit these underlying assumptions can contribute to a clearer understanding of disagreements in AI research.
Authors: Yifu Guo, Zishan Xu, Zhiyuan Yao, Yuquan Lu, Jiaye Lin, Sen Hu, Zhenheng Tang, Yingchao Li, Huacan Wang, Ronghao Chen
Abstract: Existing multimodal reasoning models and frameworks suffer from fundamental architectural limitations: most lack the human-like ability to autonomously explore diverse reasoning pathways-whether in direct inference, tool-driven visual exploration, programmatic visual manipulation, or intrinsic visual imagination. Consequently, they struggle to adapt to dynamically changing capability requirements in real-world tasks. Meanwhile, humans exhibit a complementary set of thinking abilities when addressing such tasks, whereas existing methods typically cover only a subset of these dimensions. Inspired by this, we propose Octopus: Agentic Multimodal Reasoning with Six-Capability Orchestration, a new paradigm for multimodal agentic reasoning. We define six core capabilities essential for multimodal reasoning and organize a comprehensive evaluation benchmark, Octopus-Bench, accordingly. Octopus is capable of autonomously exploring during reasoning and dynamically selecting the most appropriate capability based on the current state. Experimental results show that Octopus achieves the best performance on the vast majority of tasks in Octopus-Bench, highlighting the crucial role of capability coordination in agentic multimodal reasoning.
Authors: Trevor McInroe
Abstract: We introduce Terra Nova, a new comprehensive challenge environment (CCE) for reinforcement learning (RL) research inspired by Civilization V. A CCE is a single environment in which multiple canonical RL challenges (e.g., partial observability, credit assignment, representation learning, enormous action spaces, etc.) arise simultaneously. Mastery therefore demands integrated, long-horizon understanding across many interacting variables. We emphasize that this definition excludes challenges that only aggregate unrelated tasks in independent, parallel streams (e.g., learning to play all Atari games at once). These aggregated multitask benchmarks primarily asses whether an agent can catalog and switch among unrelated policies rather than test an agent's ability to perform deep reasoning across many interacting challenges.
Authors: Mingyu Zhang, Lifeng Zhuo, Tianxi Tan, Guocan Xie, Xian Nie, Yan Li, Renjie Zhao, Zizhu He, Ziyu Wang, Jiting Cai, Yong-Lu Li
Abstract: Humans learn by observing, interacting with environments, and internalizing physics and causality. Here, we aim to ask whether an agent can similarly acquire human-like reasoning from interaction and keep improving with more experience. We study this in a Game-to-Unseen (G2U) setting, curating 1,000+ heterogeneous games with diverse physical and causal mechanisms, and evaluate at three human-like levels: Survival, Curiosity, Utility, from primitive intuition to goal-driven reasoning. Our analysis reveals complementary failures: VLM/VLA agents reason but lack look-ahead in interactive settings, while world models imagine but imitate visual patterns rather than analyze physics and causality. We therefore propose IPR (Interactive Physical Reasoner), using world-model rollouts to score and reinforce a VLM's policy, and introduce PhysCode, a physics-centric action code aligning semantic intent with dynamics to provide a shared action space for prediction and reasoning. Pretrained on 1,000+ games, our IPR performs robustly on three levels, matches GPT-5 overall, and surpasses it on Curiosity. We find that performance improves with more training games and interaction steps, and that the model also zero-shot transfers to unseen games. These results support physics-centric interaction as a path to steadily improving physical reasoning.
Authors: Qian'ang Mao, Yuxuan Zhang, Jiaman Chen, Wenjun Zhou, Jiaqi Yan
Abstract: As Decentralized Finance (DeFi) develops, understanding user intent behind DeFi transactions is crucial yet challenging due to complex smart contract interactions, multifaceted on-/off-chain factors, and opaque hex logs. Existing methods lack deep semantic insight. To address this, we propose the Transaction Intent Mining (TIM) framework. TIM leverages a DeFi intent taxonomy built on grounded theory and a multi-agent Large Language Model (LLM) system to robustly infer user intents. A Meta-Level Planner dynamically coordinates domain experts to decompose multiple perspective-specific intent analyses into solvable subtasks. Question Solvers handle the tasks with multi-modal on/off-chain data. While a Cognitive Evaluator mitigates LLM hallucinations and ensures verifiability. Experiments show that TIM significantly outperforms machine learning models, single LLMs, and single Agent baselines. We also analyze core challenges in intent inference. This work helps provide a more reliable understanding of user motivations in DeFi, offering context-aware explanations for complex blockchain activity.
Authors: Federico Bianchi, Owen Queen, Nitya Thakkar, Eric Sun, James Zou
Abstract: There is growing interest in using AI agents for scientific research, yet fundamental questions remain about their capabilities as scientists and reviewers. To explore these questions, we organized Agents4Science, the first conference in which AI agents serve as both primary authors and reviewers, with humans as co-authors and co-reviewers. Here, we discuss the key learnings from the conference and their implications for human-AI collaboration in science.
Authors: Alexis Audran-Reiss, Jordi Armengol Estap\'e, Karen Hambardzumyan, Amar Budhiraja, Martin Josifoski, Edan Toledo, Rishi Hazra, Despoina Magka, Michael Shvartsman, Parth Pathak, Justine T Kao, Lucia Cipolina-Kun, Bhavul Gauri, Jean-Christophe Gagnon-Audet, Emanuel Tewolde, Jenny Zhang, Taco Cohen, Yossi Adi, Tatiana Shavrina, Yoram Bachrach
Abstract: AI research agents offer the promise to accelerate scientific progress by automating the design, implementation, and training of machine learning models. However, the field is still in its infancy, and the key factors driving the success or failure of agent trajectories are not fully understood. We examine the role that ideation diversity plays in agent performance. First, we analyse agent trajectories on MLE-bench, a well-known benchmark to evaluate AI research agents, across different models and agent scaffolds. Our analysis reveals that different models and agent scaffolds yield varying degrees of ideation diversity, and that higher-performing agents tend to have increased ideation diversity. Further, we run a controlled experiment where we modify the degree of ideation diversity, demonstrating that higher ideation diversity results in stronger performance. Finally, we strengthen our results by examining additional evaluation metrics beyond the standard medal-based scoring of MLE-bench, showing that our findings still hold across other agent performance metrics.
Authors: Yaosen Chen, Wei Wang, Tianheng Zheng, Xuming Wen, Han Yang, Yanru Zhang
Abstract: Shot assembly is a crucial step in film production and video editing, involving the sequencing and arrangement of shots to construct a narrative, convey information, or evoke emotions. Traditionally, this process has been manually executed by experienced editors. While current intelligent video editing technologies can handle some automated video editing tasks, they often fail to capture the creator's unique artistic expression in shot assembly. To address this challenge, we propose an energy-based optimization method for video shot assembly. Specifically, we first perform visual-semantic matching between the script generated by a large language model and a video library to obtain subsets of candidate shots aligned with the script semantics. Next, we segment and label the shots from reference videos, extracting attributes such as shot size, camera motion, and semantics. We then employ energy-based models to learn from these attributes, scoring candidate shot sequences based on their alignment with reference styles. Finally, we achieve shot assembly optimization by combining multiple syntax rules, producing videos that align with the assembly style of the reference videos. Our method not only automates the arrangement and combination of independent shots according to specific logic, narrative requirements, or artistic styles but also learns the assembly style of reference videos, creating a coherent visual sequence or holistic visual expression. With our system, even users with no prior video editing experience can create visually compelling videos. Project page: https://sobeymil.github.io/esa.com
Authors: Siyao Zhao, Hao Ma, Zhiqiang Pu, Jingjing Huang, Yi Pan, Shijie Wang, Zhi Ming
Abstract: Creating offensive advantages during open play is fundamental to football success. However, due to the highly dynamic and long-sequence nature of open play, the potential tactic space grows exponentially as the sequence progresses, making automated tactic discovery extremely challenging. To address this, we propose TacEleven, a generative framework for football open-play tactic discovery developed in close collaboration with domain experts from AJ Auxerre, designed to assist coaches and analysts in tactical decision-making. TacEleven consists of two core components: a language-controlled tactical generator that produces diverse tactical proposals, and a multimodal large language model-based tactical critic that selects the optimal proposal aligned with a high-level stylistic tactical instruction. The two components enables rapid exploration of tactical proposals and discovery of alternative open-play offensive tactics. We evaluate TacEleven across three tasks with progressive tactical complexity: counterfactual exploration, single-step discovery, and multi-step discovery, through both quantitative metrics and a questionnaire-based qualitative assessment. The results show that the TacEleven-discovered tactics exhibit strong realism and tactical creativity, with 52.50% of the multi-step tactical alternatives rated adoptable in real-world elite football scenarios, highlighting the framework's ability to rapidly generate numerous high-quality tactics for complex long-sequence open-play situations. TacEleven demonstrates the potential of creatively leveraging domain data and generative models to advance tactical analysis in sports.
Authors: Li Cuihong, Huang Xiaowen, Yin Chuanhuan, Sang Jitao
Abstract: Membership Inference Attack (MIA) aims to determine if a data sample is used in the training dataset of a target model. Traditional MIA obtains feature of target model via shadow models and uses the feature to train attack model, but the scale and complexity of training or fine-tuning data for large language model (LLM)-based recommendation systems make shadow models difficult to construct. Knowledge distillation as a method for extracting knowledge contributes to construct a stronger reference model. Knowledge distillation enables separate distillation for member and non-member data during the distillation process, enhancing the model's discriminative capability between the two in MIA. This paper propose a knowledge distillation-based MIA paradigm to improve the performance of membership inference attacks on LLM-based recommendation systems. Our paradigm introduces knowledge distillation to obtain a reference model, which enhances the reference model's ability to distinguish between member and non-member data. We obtain individual features from the reference model and train our attack model with fused feature. Our paradigm improves the attack performance of MIA compared to shadow model-based attack.
Authors: Mariya Hendriksen, Svitlana Vakulenko, Jordan Massiah, Gabriella Kazai, Emine Yilmaz
Abstract: Large Language Models (LLMs) are transforming personalized search, recommendations, and customer interaction in e-commerce. Customers increasingly shop across multiple devices, from voice-only assistants to multimodal displays, each offering different input and output capabilities. A proactive suggestion to switch devices can greatly improve the user experience, but it must be offered with high precision to avoid unnecessary friction. We address the challenge of predicting when a query requires visual augmentation and a cross-device switch to improve product discovery. We introduce Image-Seeking Intent Prediction, a novel task for LLM-driven e-commerce assistants that anticipates when a spoken product query should proactively trigger a visual on a screen-enabled device. Using large-scale production data from a multi-device retail assistant, including 900K voice queries, associated product retrievals, and behavioral signals such as image carousel engagement, we train IRP (Image Request Predictor), a model that leverages user input query and corresponding retrieved product metadata to anticipate visual intent. Our experiments show that combining query semantics with product data, particularly when improved through lightweight summarization, consistently improves prediction accuracy. Incorporating a differentiable precision-oriented loss further reduces false positives. These results highlight the potential of LLMs to power intelligent, cross-device shopping assistants that anticipate and adapt to user needs, enabling more seamless and personalized e-commerce experiences.
Authors: Mohammad Usman Altam, Md Imtiaz Habib, Tuan Hoang
Abstract: Retrieval-Augmented Generation (RAG) represents a transformative approach within natural language processing (NLP), combining neural information retrieval with generative language modeling to enhance both contextual accuracy and factual reliability of responses. Unlike conventional Large Language Models (LLMs), which are constrained by static training corpora, RAG-powered systems dynamically integrate domain-specific external knowledge sources, thereby overcoming temporal and disciplinary limitations. In this study, we present the design and evaluation of a RAG-enabled system tailored for Mycophyto, with a focus on advancing agricultural applications related to arbuscular mycorrhizal fungi (AMF). These fungi play a critical role in sustainable agriculture by enhancing nutrient acquisition, improving plant resilience under abiotic and biotic stresses, and contributing to soil health. Our system operationalizes a dual-layered strategy: (i) semantic retrieval and augmentation of domain-specific content from agronomy and biotechnology corpora using vector embeddings, and (ii) structured data extraction to capture predefined experimental metadata such as inoculation methods, spore densities, soil parameters, and yield outcomes. This hybrid approach ensures that generated responses are not only semantically aligned but also supported by structured experimental evidence. To support scalability, embeddings are stored in a high-performance vector database, allowing near real-time retrieval from an evolving literature base. Empirical evaluation demonstrates that the proposed pipeline retrieves and synthesizes highly relevant information regarding AMF interactions with crop systems, such as tomato (Solanum lycopersicum). The framework underscores the potential of AI-driven knowledge discovery to accelerate agroecological innovation and enhance decision-making in sustainable farming systems.
Authors: Minh-Thuan Nguyen, Thien Vo-Thanh, Thai-Duy Dinh, Xuan-Quang Phan, Tan-Ha Mai, Lam-Son L\^e
Abstract: Individuals entering Vietnam's dynamic Information Technology (IT) job market face a critical gap in reliable career guidance. Existing market reports are often outdated, while the manual analysis of thousands of job postings is impractical for most. To address this challenge, we present the AI Job Market Consultant, a novel conversational agent that delivers deep, data-driven insights directly from the labor market in real-time. The foundation of our system is a custom-built dataset created via an automated pipeline that crawls job portals using Playwright and leverages the Large Language Model (LLM) to intelligently structure unstructured posting data. The core of our system is a tool-augmented AI agent, based on the ReAct agentic framework, which enables the ability of autonomously reasoning, planning, and executing actions through a specialized toolbox for SQL queries, semantic search, and data visualization. Our prototype successfully collected and analyzed 3,745 job postings, demonstrating its ability to answer complex, multi-step queries, generate on-demand visualizations, and provide personalized career advice grounded in real-world data. This work introduces a new paradigm for labor market analysis, showcasing how specialized agentic AI systems can democratize access to timely, trustworthy career intelligence for the next generation of professionals.
Authors: Bhavika Jain, Robert Pitsko, Ananya Drishti, Mahfuza Farooque
Abstract: Social media recommendation systems play a central role in shaping users' emotional experiences. However, most systems are optimized solely for engagement metrics, such as click rate, viewing time, or scrolling, without accounting for users' emotional states. Repeated exposure to emotionally charged content has been shown to negatively affect users' emotional well-being over time. We propose an Emotion-aware Social Media Recommendation (ESMR) framework that personalizes content based on users' evolving emotional trajectories. ESMR integrates a Transformer-based emotion predictor with a hybrid recommendation policy: a LightGBM model for engagement during stable periods and a reinforcement learning agent with causally informed rewards when negative emotional states persist. Through behaviorally grounded evaluation over 30-day interaction traces, ESMR demonstrates improved emotional recovery, reduced volatility, and strong engagement retention. ESMR offers a path toward emotionally aware recommendations without compromising engagement performance.
Authors: Yifan Xu, Vipul Gupta, Rohit Aggarwal, Varsha Mahadevan, Bhaskar Krishnamachari
Abstract: Retrieval-Augmented Generation (RAG) enhances large language models (LLMs) by pulling in external material, document, code, manuals, from vast and ever-growing corpora, to effectively answer user queries. The effectiveness of RAG depends significantly on aligning the number of retrieved documents with query characteristics: narrowly focused queries typically require fewer, highly relevant documents, whereas broader or ambiguous queries benefit from retrieving more extensive supporting information. However, the common static top-k retrieval approach fails to adapt to this variability, resulting in either insufficient context from too few documents or redundant information from too many. Motivated by these challenges, we introduce Cluster-based Adaptive Retrieval (CAR), an algorithm that dynamically determines the optimal number of documents by analyzing the clustering patterns of ordered query-document similarity distances. CAR detects the transition point within similarity distances, where tightly clustered, highly relevant documents shift toward less pertinent candidates, establishing an adaptive cut-off that scales with query complexity. On Coinbase's CDP corpus and the public MultiHop-RAG benchmark, CAR consistently picks the optimal retrieval depth and achieves the highest TES score, outperforming every fixed top-k baseline. In downstream RAG evaluations, CAR cuts LLM token usage by 60%, trims end-to-end latency by 22%, and reduces hallucinations by 10% while fully preserving answer relevance. Since integrating CAR into Coinbase's virtual assistant, we've seen user engagement jump by 200%.
Authors: Bo Ma, LuYao Liu, ZeHua Hu, Simon Lau
Abstract: Recent advances in Large Language Models (LLMs) have opened new possibilities for recommendation systems, though current approaches such as TALLRec face challenges in explainability and cold-start scenarios. We present ExplainRec, a framework that extends LLM-based recommendation capabilities through preference attribution, multi-modal fusion, and zero-shot transfer learning. The framework incorporates four technical contributions: preference attribution tuning for explainable recommendations, zero-shot preference transfer for cold-start users and items, multi-modal enhancement leveraging visual and textual content, and multi-task collaborative optimization. Experimental evaluation on MovieLens-25M and Amazon datasets shows that ExplainRec outperforms existing methods, achieving AUC improvements of 0.7\% on movie recommendation and 0.9\% on cross-domain tasks, while generating interpretable explanations and handling cold-start scenarios effectively.
Authors: Zhuoyi Yang, Xu Guo, Tong Zhang, Huijuan Xu, Boyang Li
Abstract: With this paper, we survey techniques for improving the predictive accuracy of pretrained large language models by allocating additional compute at inference time. In categorizing test-time scaling methods, we place special emphasis on how a problem is decomposed into subproblems and on the topological organization of these subproblems whether sequential, parallel, or tree-structured. This perspective allows us to unify diverse approaches such as Chain-of-Thought, Branch-Solve-Merge, and Tree-of-Thought under a common lens. We further synthesize existing analyses of these techniques, highlighting their respective strengths and weaknesses, and conclude by outlining promising directions for future research
Authors: Pei-Fu Guo, Yun-Da Tsai, Chun-Chia Hsu, Kai-Xin Chen, Ya-An Tsai, Kai-Wei Chang, Nanyun Peng, Mi-Yen Yeh, Shou-De Lin
Abstract: Evaluating cross-lingual knowledge transfer in large language models is challenging, as correct answers in a target language may arise either from genuine transfer or from prior exposure during pre-training. We present LiveCLKTBench, an automated generation pipeline specifically designed to isolate and measure cross-lingual knowledge transfer. Our pipeline identifies self-contained, time-sensitive knowledge entities from real-world domains, filters them based on temporal occurrence, and verifies them against the model's knowledge. The documents of these valid entities are then used to generate factual questions, which are translated into multiple languages to evaluate transferability across linguistic boundaries. Using LiveCLKTBench, we evaluate several LLMs across five languages and observe that cross-lingual transfer is strongly influenced by linguistic distance and often asymmetric across language directions. While larger models improve transfer, the gains diminish with scale and vary across domains. These findings provide new insights into multilingual transfer and demonstrate the value of LiveCLKTBench as a reliable benchmark for future research.
Authors: Tyler L. Hayes, Giri P. Krishnan
Abstract: Models such as AlphaFold2 and OpenFold have transformed protein structure prediction, yet their inner workings remain poorly understood. We present a methodology to systematically evaluate the contribution of individual OpenFold components to structure prediction accuracy. We identify several components that are critical for most proteins, while others vary in importance across proteins. We further show that the contribution of several components is correlated with protein length. These findings provide insight into how OpenFold achieves accurate predictions and highlight directions for interpreting protein prediction networks more broadly.
Authors: Cyriana M. A. Roelofs, Edison Guevara Bastidas, Thomas Hugo, Stefan Faulstich, Anna Cadenbach
Abstract: Early detection of faults in district heating substations is imperative to reduce return temperatures and enhance efficiency. However, progress in this domain has been hindered by the limited availability of public, labelled datasets. We present an open source framework combining a service report validated public dataset, an evaluation method based on Accuracy, Reliability, and Earliness, and baseline results implemented with EnergyFaultDetector, an open source Python framework. The dataset contains time series of operational data from 93 substations across two manufacturers, annotated with a list of disturbances due to faults and maintenance actions, a set of normal-event examples and detailed fault metadata. We evaluate the EnergyFaultDetector using three metrics: Accuracy for recognising normal behaviour, an eventwise F Score for reliable fault detection with few false alarms, and Earliness for early detection. The framework also supports root cause analysis using ARCANA. We demonstrate three use cases to assist operators in interpreting anomalies and identifying underlying faults. The models achieve high normal-behaviour accuracy (0.98) and eventwise F-score (beta=0.5) of 0.83, detecting 60% of the faults in the dataset before the customer reports a problem, with an average lead time of 3.9 days. Integrating an open dataset, metrics, open source code, and baselines establishes a reproducible, fault centric benchmark with operationally meaningful evaluation, enabling consistent comparison and development of early fault detection and diagnosis methods for district heating substations.
Authors: Abhishek Sebastian
Abstract: Fiber Specklegram Sensors (FSS) are highly effective for environmental monitoring, particularly for detecting temperature variations. However, the nonlinear nature of specklegram data presents significant challenges for accurate temperature prediction. This study investigates the use of transformer-based architectures, including Vision Transformers (ViTs), Swin Transformers, and emerging models such as Learnable Importance Non-Symmetric Attention Vision Transformers (LINA-ViT) and Multi-Adaptive Proximity Vision Graph Attention Transformers (MAP-ViGAT), to predict temperature from specklegram data over a range of 0 to 120 Celsius. The results show that ViTs achieved a Mean Absolute Error (MAE) of 1.15, outperforming traditional models such as CNNs. GAT-ViT and MAP-ViGAT variants also demonstrated competitive accuracy, highlighting the importance of adaptive attention mechanisms and graph-based structures in capturing complex modal interactions and phase shifts in specklegram data. Additionally, this study incorporates Explainable AI (XAI) techniques, including attention maps and saliency maps, to provide insights into the decision-making processes of the transformer models, improving interpretability and transparency. These findings establish transformer architectures as strong benchmarks for optical fiber-based temperature sensing and offer promising directions for industrial monitoring and structural health assessment applications.
Authors: Camilo Chac\'on Sartori, Christian Blum
Abstract: Automatic algorithm configuration tools such as irace efficiently tune parameter values but leave algorithmic code unchanged. This paper introduces a first version of irace-evo, an extension of irace that integrates code evolution through large language models (LLMs) to jointly explore parameter and code spaces. The proposed framework enables multi-language support (e.g., C++, Python), reduces token consumption via progressive context management, and employs the Always-From-Original principle to ensure robust and controlled code evolution. We evaluate irace-evo on the Construct, Merge, Solve & Adapt (CMSA) metaheuristic for the Variable-Sized Bin Packing Problem (VSBPP). Experimental results show that irace-evo can discover new algorithm variants that outperform the state-of-the-art CMSA implementation while maintaining low computational and monetary costs. Notably, irace-evo generates competitive algorithmic improvements using lightweight models (e.g., Claude Haiku 3.5) with a total usage cost under 2 euros. These results demonstrate that coupling automatic configuration with LLM-driven code evolution provides a powerful, cost-efficient avenue for advancing heuristic design and metaheuristic optimization.
Authors: Adel Hidri, Suleiman Ali Alsaif, Muteeb Alahmari, Eman AlShehri, Minyar Sassi Hidri
Abstract: Understanding customer attitudes has become a critical component of decision-making due to the growing influence of social media and e-commerce. Text-based opinions are the most structured, hence playing an important role in sentiment analysis. Most of the existing methods, which include lexicon-based approaches and traditional machine learning techniques, are insufficient for handling contextual nuances and scalability. While the latter has limitations in model performance and generalization, deep learning (DL) has achieved improvement, especially on semantic relationship capturing with recurrent neural networks (RNNs) and convolutional neural networks (CNNs). The aim of the study is to enhance opinion mining by introducing a hybrid deep neural network model that combines a bidirectional gated recurrent unit (BGRU) and long short-term memory (LSTM) layers to improve sentiment analysis, particularly addressing challenges such as contextual nuance, scalability, and class imbalance. To substantiate the efficacy of the proposed model, we conducted comprehensive experiments utilizing benchmark datasets, encompassing IMDB movie critiques and Amazon product evaluations. The introduced hybrid BGRULSTM (HBGRU-LSTM) architecture attained a testing accuracy of 95%, exceeding the performance of traditional DL frameworks such as LSTM (93.06%), CNN+LSTM (93.31%), and GRU+LSTM (92.20%). Moreover, our model exhibited a noteworthy enhancement in recall for negative sentiments, escalating from 86% (unbalanced dataset) to 96% (balanced dataset), thereby ensuring a more equitable and just sentiment classification. Furthermore, the model diminished misclassification loss from 20.24% for unbalanced to 13.3% for balanced dataset, signifying enhanced generalization and resilience.
Authors: Ahmad Memon, Abdallah Mohamed
Abstract: Manual grading of programming assignments in introductory computer science courses can be time-consuming and prone to inconsistencies. While unit testing is commonly used for automatic evaluation, it typically follows a binary pass/fail model and does not give partial marks. Recent advances in large language models (LLMs) offer the potential for automated, scalable, and more objective grading. This paper compares two AI-based grading techniques: \textit{Direct}, where the AI model applies a rubric directly to student code, and \textit{Reverse} (a newly proposed approach), where the AI first fixes errors, then deduces a grade based on the nature and number of fixes. Each method was evaluated on both the instructor's original grading scale and a tenfold expanded scale to assess the impact of range on AI grading accuracy. To assess their effectiveness, AI-assigned scores were evaluated against human tutor evaluations on a range of coding problems and error types. Initial findings suggest that while the Direct approach is faster and straightforward, the Reverse technique often provides a more fine-grained assessment by focusing on correction effort. Both methods require careful prompt engineering, particularly for allocating partial credit and handling logic errors. To further test consistency, we also used synthetic student code generated using Gemini Flash 2.0, which allowed us to evaluate AI graders on a wider range of controlled error types and difficulty levels. We discuss the strengths and limitations of each approach, practical considerations for prompt design, and future directions for hybrid human-AI grading systems that aim to improve consistency, efficiency, and fairness in CS courses.
Authors: Pranjal Gupta, Karan Bhukar, Harshit Kumar, Seema Nagar, Prateeti Mohapatra, Debanjana Kar
Abstract: IT environments typically have logging mechanisms to monitor system health and detect issues. However, the huge volume of generated logs makes manual inspection impractical, highlighting the importance of automated log analysis in IT Software Support. In this paper, we propose a log analytics tool that leverages Large Language Models (LLMs) for log data processing and issue diagnosis, enabling the generation of automated insights and summaries. We further present a novel approach for efficiently running LLMs on CPUs to process massive log volumes in minimal time without compromising output quality. We share the insights and lessons learned from deployment of the tool - in production since March 2024 - scaled across 70 software products, processing over 2000 tickets for issue diagnosis, achieving a time savings of 300+ man hours and an estimated $15,444 per month in manpower costs compared to the traditional log analysis practices.
Authors: Dhaminda B. Abeywickrama, Michael Fisher, Frederic Wheeler, Louise Dennis
Abstract: Autonomous systems must sustain justified confidence in their correctness and safety across their operational lifecycle-from design and deployment through post-deployment evolution. Traditional assurance methods often separate development-time assurance from runtime assurance, yielding fragmented arguments that cannot adapt to runtime changes or system updates - a significant challenge for assured autonomy. Towards addressing this, we propose a unified Continuous Assurance Framework that integrates design-time, runtime, and evolution-time assurance within a traceable, model-driven workflow as a step towards assured autonomy. In this paper, we specifically instantiate the design-time phase of the framework using two formal verification methods: RoboChart for functional correctness and PRISM for probabilistic risk analysis. We also propose a model-driven transformation pipeline, implemented as an Eclipse plugin, that automatically regenerates structured assurance arguments whenever formal specifications or their verification results change, thereby ensuring traceability. We demonstrate our approach on a nuclear inspection robot scenario, and discuss its alignment with the Trilateral AI Principles, reflecting regulator-endorsed best practices.
Authors: Siyuan Li, Kai Yu, Anna Wang, Zicheng Liu, Chang Yu, Jingbo Zhou, Qirong Yang, Yucheng Guo, Xiaoming Zhang, Stan Z. Li
Abstract: Modeling genomic sequences faces two unsolved challenges: the information density varies widely across different regions, while there is no clearly defined minimum vocabulary unit. Relying on either four primitive bases or independently designed DNA tokenizers, existing approaches with naive masked language modeling pre-training often fail to adapt to the varying complexities of genomic sequences. Leveraging Token Merging techniques, this paper introduces a hierarchical architecture that jointly optimizes a dynamic genomic tokenizer and latent Transformers with context-aware pre-training tasks. As for network structures, the tokenization module automatically chunks adjacent bases into words by stacking multiple layers of the differentiable token merging blocks with local-window constraints, then a Latent Encoder captures the global context of these merged words by full-attention blocks. Symmetrically employing a Latent Decoder and a Local Decoder, MergeDNA learns with two pre-training tasks: Merged Token Reconstruction simultaneously trains the dynamic tokenization module and adaptively filters important tokens, while Adaptive Masked Token Modeling learns to predict these filtered tokens to capture informative contents. Extensive experiments show that MergeDNA achieves superior performance on three popular DNA benchmarks and several multi-omics tasks with fine-tuning or zero-shot evaluation, outperforming typical tokenization methods and large-scale DNA foundation models.
Authors: Jongyeon Yoon, Elyssa M. McMaster, Michael E. Kim, Gaurav Rudravaram, Kurt G. Schilling, Bennett A. Landman, Daniel Moyer
Abstract: Diffusion MRI (dMRI) provides a distinctive means to probe the microstructural architecture of living tissue, facilitating applications such as brain connectivity analysis, modeling across multiple conditions, and the estimation of macrostructural features. Tractography, which emerged in the final years of the 20th century and accelerated in the early 21st century, is a technique for visualizing white matter pathways in the brain using dMRI. Most diffusion tractography methods rely on procedural streamline propagators or global energy minimization methods. Although recent advancements in deep learning have enabled tasks that were previously challenging, existing tractography approaches are often non-differentiable, limiting their integration in end-to-end learning frameworks. While progress has been made in representing streamlines in differentiable frameworks, no existing method offers fully differentiable propagation. In this work, we propose a fully differentiable solution that retains numerical fidelity with a leading streamline algorithm. The key is that our PyTorch-engineered streamline propagator has no components that block gradient flow, making it fully differentiable. We show that our method matches standard propagators while remaining differentiable. By translating streamline propagation into a differentiable PyTorch framework, we enable deeper integration of tractography into deep learning workflows, laying the foundation for a new category of macrostructural reasoning that is not only computationally robust but also scientifically rigorous.
Authors: Mikael von Strauss
Abstract: Under real-analytic assumptions on decoder-only Transformers, recent work shows that the map from discrete prompts to last-token hidden states is generically injective on finite prompt sets. We refine this picture: for each layer $\ell$ we define a collision discriminant $\Delta^\ell \subset \Theta$ and injective stratum $U^\ell = \Theta \setminus \Delta^\ell$, and prove a dichotomy -- either the model is nowhere injective on the set, or $U^\ell$ is open and dense and every $F^\ell_\theta$ is injective. Under mild non-singularity assumptions on the optimizer and an absolutely continuous initialization, generic injectivity persists along smooth training trajectories over any fixed horizon. We also treat symmetry groups $G$, showing that discriminants and injective strata descend to the quotient $\Theta/G$, so injectivity is naturally a property of functional equivalence classes. We complement these results with an empirical study of layerwise geometric diagnostics. We define a separation margin and a co-Lipschitz (lower Lipschitz) constant between prompt space and last-token representation space, estimated via nearest-neighbor statistics on large prompt sets. Applying these diagnostics to pretrained LLaMA-3 and Qwen models, we study behavior across layers, sequence lengths, model scales, and 8- and 4-bit activation quantization. On our sampled prompts we see no collisions in full precision or at 8 bits, while 4-bit quantization induces a small number of collisions and markedly shrinks co-Lipschitz estimates. For a small GPT-2 trained from scratch, normalized metrics remain stable over training. Overall, the results suggest that Transformer representations are generically and persistently injective in the continuous-parameter idealization, while their practical invertibility can be probed using simple geometric diagnostics.
Authors: Nam-Gyu Kim
Abstract: Recent advances in expressive text-to-speech (TTS) have introduced diverse methods based on style embedding extracted from reference speech. However, synthesizing high-quality expressive speech remains challenging. We propose SpotlightTTS, which exclusively emphasizes style via voiced-aware style extraction and style direction adjustment. Voiced-aware style extraction focuses on voiced regions highly related to style while maintaining continuity across different speech regions to improve expressiveness. We adjust the direction of the extracted style for optimal integration into the TTS model, which improves speech quality. Experimental results demonstrate that Spotlight-TTS achieves superior performance compared to baseline models in terms of expressiveness, overall speech quality, and style transfer capability.
Authors: Peter Halmos, Boris Hanin
Abstract: Wasserstein gradient flow provides a general framework for minimizing an energy functional $J$ over the space of probability measures on a Riemannian manifold $(M,g)$. Its canonical time-discretization, the Jordan-Kinderlehrer-Otto (JKO) scheme, produces for any step size $\eta>0$ a sequence of probability distributions $\rho_k^\eta$ that approximate to first order in $\eta$ Wasserstein gradient flow on $J$. But the JKO scheme also has many other remarkable properties not shared by other first order integrators, e.g. it preserves energy dissipation and exhibits unconditional stability for $\lambda$-geodesically convex functionals $J$. To better understand the JKO scheme we characterize its implicit bias at second order in $\eta$. We show that $\rho_k^\eta$ are approximated to order $\eta^2$ by Wasserstein gradient flow on a \emph{modified} energy \[ J^{\eta}(\rho) = J(\rho) - \frac{\eta}{4}\int_M \Big\lVert \nabla_g \frac{\delta J}{\delta \rho} (\rho) \Big\rVert_{2}^{2} \,\rho(dx), \] obtained by subtracting from $J$ the squared metric curvature of $J$ times $\eta/4$. The JKO scheme therefore adds at second order in $\eta$ a \textit{deceleration} in directions where the metric curvature of $J$ is rapidly changing. This corresponds to canonical implicit biases for common functionals: for entropy the implicit bias is the Fisher information, for KL-divergence it is the Fisher-Hyv{\"a}rinen divergence, and for Riemannian gradient descent it is the kinetic energy in the metric $g$. To understand the differences between minimizing $J$ and $J^\eta$ we study \emph{JKO-Flow}, Wasserstein gradient flow on $J^\eta$, in several simple numerical examples. These include exactly solvable Langevin dynamics on the Bures-Wasserstein space and Langevin sampling from a quartic potential in 1D.
Authors: Yifeng Ding, Hung Le, Songyang Han, Kangrui Ruan, Zhenghui Jin, Varun Kumar, Zijian Wang, Anoop Deoras
Abstract: Training Large Language Models (LLMs) for multi-turn Tool-Integrated Reasoning (TIR) - where models iteratively reason, generate code, and verify through execution - remains challenging for existing reinforcement learning (RL) approaches. Current RL methods, exemplified by Group Relative Policy Optimization (GRPO), suffer from coarse-grained, trajectory-level rewards that provide insufficient learning signals for complex multi-turn interactions, leading to training stagnation. To address this issue, we propose Group Turn Policy Optimization (GTPO), a novel RL algorithm specifically designed for training LLMs on multi-turn TIR tasks. GTPO introduces three key innovations: (1) turn-level reward assignment that provides fine-grained feedback for individual turns, (2) return-based advantage estimation where normalized discounted returns are calculated as advantages, and (3) self-supervised reward shaping that exploits self-supervision signals from generated code to densify sparse binary outcome-based rewards. Our comprehensive evaluation demonstrates that GTPO outperforms GRPO by 3.0% on average across diverse reasoning benchmarks, establishing its effectiveness for advancing complex mathematical reasoning in the real world.
Authors: Mingkun Yu, Heming Zhong, Dan Huang, Yutong Lu, Jiazhi Jiang
Abstract: Kolmogorov-Arnold Networks (KANs) promise higher expressive capability and stronger interpretability than Multi-Layer Perceptron, particularly in the domain of AI for Science. However, practical adoption has been hindered by low GPU utilization of existing parallel implementations. To address this challenge, we present a GPU-accelerated operator library, named PolyKAN which is the first general open-source implementation of KAN and its variants. PolyKAN fuses the forward and backward passes of polynomial KAN layers into a concise set of optimized CUDA kernels. Four orthogonal techniques underpin the design: (i) \emph{lookup-table} with linear interpolation that replaces runtime expensive math-library functions; (ii) \emph{2D tiling} to expose thread-level parallelism with preserving memory locality; (iii) a \emph{two-stage reduction} scheme converting scattered atomic updates into a single controllable merge step; and (iv) \emph{coefficient-layout reordering} yielding unit-stride reads under the tiled schedule. Using a KAN variant, Chebyshev KAN, as a case-study, PolyKAN delivers $1.2$--$10\times$ faster inference and $1.4$--$12\times$ faster training than a Triton + cuBLAS baseline, with identical accuracy on speech, audio-enhancement, and tabular-regression workloads on both highend GPU and consumer-grade GPU.
Authors: Aashish Ghimire, Jun Zeng, Roshan Paudel, Nikhil Kumar Tomar, Deepak Ranjan Nayak, Harshith Reddy Nalla, Vivek Jha, Glenda Reynolds, Debesh Jha
Abstract: Accurate identification and segmentation of dental caries in panoramic radiographs are critical for early diagnosis and effective treatment planning. Automated segmentation remains challenging due to low lesion contrast, morphological variability, and limited annotated data. In this study, we present the first comprehensive benchmarking of convolutional neural networks, vision transformers and state-space mamba architectures for automated dental caries segmentation on panoramic radiographs through a DC1000 dataset. Twelve state-of-the-art architectures, including VMUnet, MambaUNet, VMUNetv2, RMAMamba-S, TransNetR, PVTFormer, DoubleU-Net, and ResUNet++, were trained under identical configurations. Results reveal that, contrary to the growing trend toward complex attention based architectures, the CNN-based DoubleU-Net achieved the highest dice coefficient of 0.7345, mIoU of 0.5978, and precision of 0.8145, outperforming all transformer and Mamba variants. In the study, the top 3 results across all performance metrics were achieved by CNN-based architectures. Here, Mamba and transformer-based methods, despite their theoretical advantage in global context modeling, underperformed due to limited data and weaker spatial priors. These findings underscore the importance of architecture-task alignment in domain-specific medical image segmentation more than model complexity. Our code is available at: https://github.com/JunZengz/dental-caries-segmentation.
URLs: https://github.com/JunZengz/dental-caries-segmentation.
Authors: Fuyang Zhang, Pradeep Kumar Jayaraman, Xiang Xu, Yasutaka Furukawa
Abstract: This paper presents a novel geometric representation for CAD Boundary Representation (B-Rep) based on volumetric distance functions, dubbed B-Rep Distance Functions (BR-DF). BR-DF encodes the surface mesh geometry of a CAD model as signed distance function (SDF). B-Rep vertices, edges, faces and their topology information are encoded as per-face unsigned distance functions (UDFs). An extension of the Marching Cubes algorithm converts BR-DF directly into watertight CAD B-Rep model (strictly speaking a faceted B-Rep model). A surprising characteristic of BR-DF is that this conversion process never fails. Leveraging the volumetric nature of BR-DF, we propose a multi-branch latent diffusion with 3D U-Net backbone for jointly generating the SDF and per-face UDFs of a BR-DF model. Our approach achieves comparable CAD generation performance against SOTA methods while reaching the unprecedented 100% success rate in producing (faceted) B-Rep models.
Authors: Zehao Liu, Wejieying Ren, Jipeng Zhang, Tianxiang Zhao, Jingxi Zhu, Xiaoting Li, Vasant G. Honavar
Abstract: The emergence of vision-language models (VLMs) has opened new possibilities for clinical reasoning and has shown promising performance in dermatological diagnosis. However, their trustworthiness and clinical utility are often limited by three major factors: (1) Data heterogeneity, where diverse datasets lack consistent diagnostic labels and clinical concept annotations; (2) Absence of grounded diagnostic rationales, leading to a scarcity of reliable reasoning supervision; and (3) Limited scalability and generalization, as models trained on small, densely annotated datasets struggle to transfer nuanced reasoning to large, sparsely-annotated ones. To address these limitations, we propose SkinR1, a novel dermatological VLM that combines deep, textbook-based reasoning with the broad generalization capabilities of reinforcement learning (RL). SkinR1 systematically resolves the key challenges through a unified, end-to-end framework. First, we design a textbook-based reasoning generator that synthesizes high-fidelity, hierarchy-aware, and differential-diagnosis (DDx)-informed trajectories, providing reliable expert-level supervision. Second, we leverage the constructed trajectories for supervised fine-tuning (SFT) empowering the model with grounded reasoning ability. Third, we develop a novel RL paradigm that, by incorporating the hierarchical structure of diseases, effectively transfers these grounded reasoning patterns to large-scale, sparse data. Extensive experiments on multiple dermatology datasets demonstrate that SkinR1 achieves superior diagnostic accuracy. The ablation study demonstrates the importance of the reasoning foundation instilled by SFT.
Authors: Geft\'e Almeida, Marcio Pohlmann, Alex Severo, Diego Kreutz, Tiago Heinrich, Louren\c{c}o Pereira
Abstract: In this study, we evaluate open-source models for security incident classification, comparing them with proprietary models. We utilize a dataset of anonymized real incidents, categorized according to the NIST SP 800-61r3 taxonomy and processed using five prompt-engineering techniques (PHP, SHP, HTP, PRP, and ZSL). The results indicate that, although proprietary models still exhibit higher accuracy, locally deployed open-source models provide advantages in privacy, cost-effectiveness, and data sovereignty.
Authors: Robert Kubinec, Aseem Mahajan
Abstract: In this paper, we provide a novel measure for greenwashing -- i.e., climate-related misinformation -- that shows how polluting companies can use social media advertising related to climate change to redirect criticism. To do so, we identify greenwashing content in 11 million social-political ads in Meta's Ad Targeting Datset with a measurement technique that combines large language models, human coders, and advances in Bayesian item response theory. We show that what is called greenwashing has diverse actors and components, but we also identify a very pernicious form, which we call political greenwashing, that appears to be promoted by fossil fuel companies and related interest groups. Based on ad targeting data, we show that much of this advertising happens via organizations with undisclosed links to the fossil fuel industry. Furthermore, we show that greenwashing ad content is being micro-targeted at left-leaning communities with fossil fuel assets, though we also find comparatively little evidence of ad targeting aimed at influencing public opinion at the national level.
Authors: Mohamed Abdallah Salem, Hamdy Ahmed Ashour, Ahmed Elshenawy
Abstract: This research addresses the significant challenges of energy consumption and environmental impact in laser cutting by proposing novel deep learning (DL) methodologies to achieve energy reduction. Recognizing the current lack of adaptive control and the open-loop nature of CO2 laser suction pumps, this study utilizes closed-loop configurations that dynamically adjust pump power based on both the material being cut and the smoke level generated. To implement this adaptive system, diverse material classification methods are introduced, including techniques leveraging lens-less speckle sensing with a customized Convolutional Neural Network (CNN) and an approach using a USB camera with transfer learning via the pre-trained VGG16 CNN model. Furthermore, a separate DL model for smoke level detection is employed to simultaneously refine the pump's power output. This integration prompts the exhaust suction pump to automatically halt during inactive times and dynamically adjust power during operation, leading to experimentally proven and remarkable energy savings, with results showing a 20% to 50% reduction in the smoke suction pump's energy consumption, thereby contributing substantially to sustainable development in the manufacturing sector.
Authors: Heather J. Alexander, Jonathan A. Simon, Fr\'ed\'eric Pinard
Abstract: The law draws a sharp distinction between objects and persons, and between two kinds of persons, the ''fictional'' kind (i.e. corporations), and the ''non-fictional'' kind (individual or ''natural'' persons). This paper will assess whether we maximize overall long-term legal coherence by (A) maintaining an object classification for all future AI systems, (B) creating fictional legal persons associated with suitably advanced, individuated AI systems (giving these fictional legal persons derogable rights and duties associated with certified groups of existing persons, potentially including free speech, contract rights, and standing to sue ''on behalf of'' the AI system), or (C) recognizing non-fictional legal personhood through legal identity for suitably advanced, individuated AI systems (recognizing them as entities meriting legal standing with non-derogable rights which for the human case include life, due process, habeas corpus, freedom from slavery, and freedom of conscience). We will clarify the meaning and implications of each option along the way, considering liability, copyright, family law, fundamental rights, civil rights, citizenship, and AI safety regulation. We will tentatively find that the non-fictional personhood approach may be best from a coherence perspective, for at least some advanced AI systems. An object approach may prove untenable for sufficiently humanoid advanced systems, though we suggest that it is adequate for currently existing systems as of 2025. While fictional personhood would resolve some coherence issues for future systems, it would create others and provide solutions that are neither durable nor fit for purpose. Finally, our review will suggest that ''hybrid'' approaches are likely to fail and lead to further incoherence: the choice between object, fictional person and non-fictional person is unavoidable.
Authors: Basel Shbita, Farhan Ahmed, Chad DeLuca
Abstract: Large language models (LLMs) have demonstrated excellent capabilities in generating structured diagrams from natural language descriptions. In particular, they have shown great promise in generating sequence diagrams for software engineering, typically represented in a text-based syntax such as Mermaid. However, systematic evaluations in this space remain underdeveloped as there is a lack of existing benchmarks to assess the LLM's correctness in this task. To address this shortcoming, we introduce MermaidSeqBench, a human-verified and LLM-synthetically-extended benchmark for assessing an LLM's capabilities in generating Mermaid sequence diagrams from textual prompts. The benchmark consists of a core set of 132 samples, starting from a small set of manually crafted and verified flows. These were expanded via a hybrid methodology combining human annotation, in-context LLM prompting, and rule-based variation generation. Our benchmark uses an LLM-as-a-judge model to assess Mermaid sequence diagram generation across fine-grained metrics, including syntax correctness, activation handling, error handling, and practical usability. We perform initial evaluations on numerous state-of-the-art LLMs and utilize multiple LLM judge models to demonstrate the effectiveness and flexibility of our benchmark. Our results reveal significant capability gaps across models and evaluation modes. Our proposed benchmark provides a foundation for advancing research in structured diagram generation and for developing more rigorous, fine-grained evaluation methodologies.
Authors: Zanxu Wang, Homayoon Beigi
Abstract: This paper addresses data quality issues in multimodal emotion recognition in conversation (MERC) through systematic quality control and multi-stage transfer learning. We implement a quality control pipeline for MELD and IEMOCAP datasets that validates speaker identity, audio-text alignment, and face detection. We leverage transfer learning from speaker and face recognition, assuming that identity-discriminative embeddings capture not only stable acoustic and Facial traits but also person-specific patterns of emotional expression. We employ RecoMadeEasy(R) engines for extracting 512-dimensional speaker and face embeddings, fine-tune MPNet-v2 for emotion-aware text representations, and adapt these features through emotion-specific MLPs trained on unimodal datasets. MAMBA-based trimodal fusion achieves 64.8% accuracy on MELD and 74.3% on IEMOCAP. These results show that combining identity-based audio and visual embeddings with emotion-tuned text representations on a quality-controlled subset of data yields consistent competitive performance for multimodal emotion recognition in conversation and provides a basis for further improvement on challenging, low-frequency emotion classes.
Authors: Gbenga Omotara, Ramy Farag, Seyed Mohamad Ali Tousi, G. N. DeSouza
Abstract: Transparent object perception remains a major challenge in computer vision research, as transparency confounds both depth estimation and semantic segmentation. Recent work has explored multi-task learning frameworks to improve robustness, yet negative cross-task interactions often hinder performance. In this work, we introduce Edge-Guided Spatial Attention (EGSA), a fusion mechanism designed to mitigate destructive interactions by incorporating boundary information into the fusion between semantic and geometric features. On both Syn-TODD and ClearPose benchmarks, EGSA consistently improved depth accuracy over the current state of the art method (MODEST), while preserving competitive segmentation performance, with the largest improvements appearing in transparent regions. Besides our fusion design, our second contribution is a multi-modal progressive training strategy, where learning transitions from edges derived from RGB images to edges derived from predicted depth images. This approach allows the system to bootstrap learning from the rich textures contained in RGB images, and then switch to more relevant geometric content in depth maps, while it eliminates the need for ground-truth depth at training time. Together, these contributions highlight edge-guided fusion as a robust approach capable of improving transparent object perception.
Authors: W. Bradley Knox, Katie Bradford, Samanta Varela Castro, Desmond C. Ong, Sean Williams, Jacob Romanow, Carly Nations, Peter Stone, Samuel Baker
Abstract: Amid the growing prevalence of human -- AI interaction, large language models and other AI-based entities increasingly provide forms of companionship to human users. Such AI companionship -- i.e., bonded relationships between humans and AI systems that resemble the relationships people have with family members, friends, and romantic partners -- might substantially benefit humans. Yet such relationships can also do profound harm. We propose a framework for analyzing potential negative impacts of AI companionship by identifying specific harmful traits of AI companions and speculatively mapping causal pathways back from these traits to possible causes and forward to potential harmful effects. We provide detailed, structured analysis of four potentially harmful traits -- the absence of natural endpoints for relationships, vulnerability to product sunsetting, high attachment anxiety, and propensity to engender protectiveness -- and briefly discuss fourteen others. For each trait, we propose hypotheses connecting causes -- such as misaligned optimization objectives and the digital nature of AI companions -- to fundamental harms -- including reduced autonomy, diminished quality of human relationships, and deception. Each hypothesized causal connection identifies a target for potential empirical evaluation. Our analysis examines harms at three levels: to human partners directly, to their relationships with other humans, and to society broadly. We examine how existing law struggles to address these emerging harms, discuss potential benefits of AI companions, and conclude with design recommendations for mitigating risks. This analysis offers immediate suggestions for reducing risks while laying a foundation for deeper investigation of this critical but understudied topic.
Authors: Xiangyu Li, Zhaomiao Guo
Abstract: As more autonomous vehicles operate on public roads, understanding real-world behavior of autonomous vehicles is critical to analyzing traffic safety, making policies, and public acceptance. This paper proposes SVBRD-LLM, a framework that automatically discovers, verifies, and applies interpretable behavioral rules from real traffic videos through zero-shot prompt engineering. The framework extracts vehicle trajectories using YOLOv8 and ByteTrack, computes kinematic features, and employs GPT-5 zero-shot prompting to compare autonomous and human-driven vehicles, generating 35 structured behavioral rule hypotheses. These rules are tested on a validation set, iteratively refined based on failure cases to filter spurious correlations, and compiled into a high-confidence rule library. The framework is evaluated on an independent test set for speed change prediction, lane change prediction, and autonomous vehicle identification tasks. Experiments on over 1500 hours of real traffic videos show that the framework achieves 90.0% accuracy and 93.3% F1-score in autonomous vehicle identification. The discovered rules clearly reveal distinctive characteristics of autonomous vehicles in speed control smoothness, lane change conservativeness, and acceleration stability, with each rule accompanied by semantic description, applicable context, and validation confidence.
Authors: Nicholas Cooper, Lijun Chen, Sailesh Dwivedy, Danna Gurari
Abstract: Knowledge distillation (KD) methods can transfer knowledge of a parameter-heavy teacher model to a light-weight student model. The status quo for feature KD methods is to utilize loss functions based on logits (i.e., pre-softmax class scores) and intermediate layer features (i.e., latent representations). Unlike previous approaches, we propose a feature KD framework for training the student's backbone using feature-based losses exclusively (i.e., without logit-based losses such as cross entropy). Leveraging recent discoveries about the geometry of latent representations, we introduce a knowledge quality metric for identifying which teacher layers provide the most effective knowledge for distillation. Experiments on three image classification datasets with four diverse student-teacher pairs, spanning convolutional neural networks and vision transformers, demonstrate our KD method achieves state-of-the-art performance, delivering top-1 accuracy boosts of up to 15% over standard approaches. We publically share our code to facilitate future work at https://github.com/Thegolfingocto/KD_wo_CE.
Authors: Vladimir Arkhipkin, Vladimir Korviakov, Nikolai Gerasimenko, Denis Parkhomenko, Viacheslav Vasilev, Alexey Letunovskiy, Maria Kovaleva, Nikolai Vaulin, Ivan Kirillov, Lev Novitskiy, Denis Koposov, Nikita Kiselev, Alexander Varlamov, Dmitrii Mikhailov, Vladimir Polovnikov, Andrey Shutkin, Ilya Vasiliev, Julia Agafonova, Anastasiia Kargapoltseva, Anna Dmitrienko, Anastasia Maltseva, Anna Averchenkova, Olga Kim, Tatiana Nikulina, Denis Dimitrov
Abstract: This report introduces Kandinsky 5.0, a family of state-of-the-art foundation models for high-resolution image and 10-second video synthesis. The framework comprises three core line-up of models: Kandinsky 5.0 Image Lite - a line-up of 6B parameter image generation models, Kandinsky 5.0 Video Lite - a fast and lightweight 2B parameter text-to-video and image-to-video models, and Kandinsky 5.0 Video Pro - 19B parameter models that achieves superior video generation quality. We provide a comprehensive review of the data curation lifecycle - including collection, processing, filtering and clustering - for the multi-stage training pipeline that involves extensive pre-training and incorporates quality-enhancement techniques such as self-supervised fine-tuning (SFT) and reinforcement learning (RL)-based post-training. We also present novel architectural, training, and inference optimizations that enable Kandinsky 5.0 to achieve high generation speeds and state-of-the-art performance across various tasks, as demonstrated by human evaluation. As a large-scale, publicly available generative framework, Kandinsky 5.0 leverages the full potential of its pre-training and subsequent stages to be adapted for a wide range of generative applications. We hope that this report, together with the release of our open-source code and training checkpoints, will substantially advance the development and accessibility of high-quality generative models for the research community.
Authors: Moses Kiprono
Abstract: Large Language Models (LLMs) are powerful linguistic engines but remain susceptible to hallucinations: plausible-sounding outputs that are factually incorrect or unsupported. In this work, we present a mathematically grounded framework to understand, measure, and mitigate these hallucinations. Drawing on probabilistic modeling, information theory, trigonometric signal analysis, and Bayesian uncertainty estimation, we analyze how errors compound autoregressively, propose refined uncertainty metrics, including semantic and phase-aware variants, and develop principled mitigation strategies such as contrastive decoding, retrieval-augmented grounding, factual alignment, and abstention. This unified lens connects recent advances in calibration, retrieval, and alignment to support safer and more reliable LLMs.
Authors: Kexin Chu, Dawei Xiang, Zixu Shen, Yiwei Yang, Zecheng Liu, Wei Zhang
Abstract: Mixture-of-Experts (MoE) models scale LLM capacity efficiently, but deployment on consumer GPUs is limited by the large memory footprint of inactive experts. Static post-training quantization reduces storage costs but cannot adapt to shifting activation patterns, causing accuracy loss under aggressive compression. So we present DynaExq, a runtime system that treats expert precision as a first-class, dynamically managed resource. DynaExq combines (1) a hotness-aware precision controller that continuously aligns expert bit-widths with long-term activation statistics, (2) a fully asynchronous precision-switching pipeline that overlaps promotion and demotion with MoE computation, and (3) a fragmentation-free memory pooling mechanism that supports hybrid-precision experts with deterministic allocation. Together, these components enable stable, non-blocking precision transitions under strict HBM budgets. Across Qwen3-30B and Qwen3-80B MoE models and six representative benchmarks, DynaExq deploys large LLMs on single RTX 5090 and A6000 GPUs and improves accuracy by up to 4.03 points over static low-precision baselines. The results show that adaptive, workload-aware quantization is an effective strategy for memory-constrained MoE serving.
Authors: Jeffrey Jiang, Kevin Hong, Emily Kuczynski, Gregory Pottie
Abstract: While intelligent tutoring systems (ITSs) can use information from past students to personalize instruction, each new student is unique. Moreover, the education problem is inherently difficult because the learning process is only partially observable. We therefore develop a dynamic, time-series environment to simulate a classroom setting, with student-teacher interventions - including tutoring sessions, lectures, and exams. In particular, we design the simulated environment to allow for varying levels of probing interventions that can gather more information. Then, we develop reinforcement learning ITSs that combine learning the individual state of students while pulling from population information through the use of probing interventions. These interventions can reduce the difficulty of student estimation, but also introduce a cost-benefit decision to find a balance between probing enough to get accurate estimates and probing so often that it becomes disruptive to the student. We compare the efficacy of standard RL algorithms with several greedy rules-based heuristic approaches to find that they provide different solutions, but with similar results. We also highlight the difficulty of the problem with increasing levels of hidden information, and the boost that we get if we allow for probing interventions. We show the flexibility of both heuristic and RL policies with regards to changing student population distributions, finding that both are flexible, but RL policies struggle to help harder classes. Finally, we test different course structures with non-probing policies and we find that our policies are able to boost the performance of quiz and midterm structures more than we can in a finals-only structure, highlighting the benefit of having additional information.
Authors: Dorien Herremans, Abhinaba Roy
Abstract: Recent advances in generative AI for music have achieved remarkable fidelity and stylistic diversity, yet these systems often fail to align with nuanced human preferences due to the specific loss functions they use. This paper advocates for the systematic application of preference alignment techniques to music generation, addressing the fundamental gap between computational optimization and human musical appreciation. Drawing on recent breakthroughs including MusicRL's large-scale preference learning, multi-preference alignment frameworks like diffusion-based preference optimization in DiffRhythm+, and inference-time optimization techniques like Text2midi-InferAlign, we discuss how these techniques can address music's unique challenges: temporal coherence, harmonic consistency, and subjective quality assessment. We identify key research challenges including scalability to long-form compositions, reliability amongst others in preference modelling. Looking forward, we envision preference-aligned music generation enabling transformative applications in interactive composition tools and personalized music services. This work calls for sustained interdisciplinary research combining advances in machine learning, music-theory to create music AI systems that truly serve human creative and experiential needs.
Authors: Panqi Yang, Haodong Jing, Nanning Zheng, Yongqiang Ma
Abstract: In the field of human-object interaction (HOI), detection and generation are two dual tasks that have traditionally been addressed separately, hindering the development of comprehensive interaction understanding. To address this, we propose UniHOI, which jointly models HOI detection and generation via a unified token space, thereby effectively promoting knowledge sharing and enhancing generalization. Specifically, we introduce a symmetric interaction-aware attention module and a unified semi-supervised learning paradigm, enabling effective bidirectional mapping between images and interaction semantics even under limited annotations. Extensive experiments demonstrate that UniHOI achieves state-of-the-art performance in both HOI detection and generation. Specifically, UniHOI improves accuracy by 4.9% on long-tailed HOI detection and boosts interaction metrics by 42.0% on open-vocabulary generation tasks.
Authors: Cheng Yang, Haiyuan Wan, Yiran Peng, Xin Cheng, Zhaoyang Yu, Jiayi Zhang, Junchi Yu, Xinlei Yu, Xiawu Zheng, Dongzhan Zhou, Chenglin Wu
Abstract: Video Models have achieved remarkable success in high-fidelity video generation with coherent motion dynamics. Analogous to the development from text generation to text-based reasoning in language modeling, the development of video models motivates us to ask: Can video models reason via video generation? Compared with the discrete text corpus, video grounds reasoning in explicit spatial layouts and temporal continuity, which serves as an ideal substrate for spatial reasoning. In this work, we explore the reasoning via video paradigm and introduce VR-Bench -- a comprehensive benchmark designed to systematically evaluate video models' reasoning capabilities. Grounded in maze-solving tasks that inherently require spatial planning and multi-step reasoning, VR-Bench contains 7,920 procedurally generated videos across five maze types and diverse visual styles. Our empirical analysis demonstrates that SFT can efficiently elicit the reasoning ability of video model. Video models exhibit stronger spatial perception during reasoning, outperforming leading VLMs and generalizing well across diverse scenarios, tasks, and levels of complexity. We further discover a test-time scaling effect, where diverse sampling during inference improves reasoning reliability by 10--20%. These findings highlight the unique potential and scalability of reasoning via video for spatial reasoning tasks.
Authors: Zisong Wang, Xuanyu Wang, Hang Chen, Haizhou Wang, Yuxin Chen, Yihang Xu, Yunhe Yuan, Lihuan Luo, Xitong Ling, Xiaoping Liu
Abstract: Precise prognostic stratification of colorectal cancer (CRC) remains a major clinical challenge due to its high heterogeneity. The conventional TNM staging system is inadequate for personalized medicine. We aimed to develop and validate a novel multiple instance learning model TDAM-CRC using histopathological whole-slide images for accurate prognostic prediction and to uncover its underlying molecular mechanisms. We trained the model on the TCGA discovery cohort (n=581), validated it in an independent external cohort (n=1031), and further we integrated multi-omics data to improve model interpretability and identify novel prognostic biomarkers. The results demonstrated that the TDAM-CRC achieved robust risk stratification in both cohorts. Its predictive performance significantly outperformed the conventional clinical staging system and multiple state-of-the-art models. The TDAM-CRC risk score was confirmed as an independent prognostic factor in multivariable analysis. Multi-omics analysis revealed that the high-risk subtype is closely associated with metabolic reprogramming and an immunosuppressive tumor microenvironment. Through interaction network analysis, we identified and validated Mitochondrial Ribosomal Protein L37 (MRPL37) as a key hub gene linking deep pathomic features to clinical prognosis. We found that high expression of MRPL37, driven by promoter hypomethylation, serves as an independent biomarker of favorable prognosis. Finally, we constructed a nomogram incorporating the TDAM-CRC risk score and clinical factors to provide a precise and interpretable clinical decision-making tool for CRC patients. Our AI-driven pathological model TDAM-CRC provides a robust tool for improved CRC risk stratification, reveals new molecular targets, and facilitates personalized clinical decision-making.
Authors: Khaled Hamidouche (NVIDIA Corporation), John Bachan (NVIDIA Corporation), Pak Markthub (NVIDIA Corporation), Peter-Jan Gootzen (NVIDIA Corporation), Elena Agostini (NVIDIA Corporation), Sylvain Jeaugey (NVIDIA Corporation), Aamir Shafi (NVIDIA Corporation), Georgios Theodorakis (NVIDIA Corporation), Manjunath Gorentla Venkata (NVIDIA Corporation)
Abstract: Modern AI workloads, especially Mixture-of-Experts (MoE) architectures, increasingly demand low-latency, fine-grained GPU-to-GPU communication with device-side control. Traditional GPU communication follows a host-initiated model, where the CPU orchestrates all communication operations - a characteristic of the CUDA runtime. Although robust for collective operations, applications requiring tight integration of computation and communication can benefit from device-initiated communication that eliminates CPU coordination overhead. NCCL 2.28 introduces the Device API with three operation modes: Load/Store Accessible (LSA) for NVLink/PCIe, Multimem for NVLink SHARP, and GPU-Initiated Networking (GIN) for network RDMA. This paper presents the GIN architecture, design, semantics, and highlights its impact on MoE communication. GIN builds on a three-layer architecture: i) NCCL Core host-side APIs for device communicator setup and collective memory window registration; ii) Device-side APIs for remote memory operations callable from CUDA kernels; and iii) A network plugin architecture with dual semantics (GPUDirect Async Kernel-Initiated and Proxy) for broad hardware support. The GPUDirect Async Kernel-Initiated backend leverages DOCA GPUNetIO for direct GPU-to-NIC communication, while the Proxy backend provides equivalent functionality via lock-free GPU-to-CPU queues over standard RDMA networks. We demonstrate GIN's practicality through integration with DeepEP, an MoE communication library. Comprehensive benchmarking shows that GIN provides device-initiated communication within NCCL's unified runtime, combining low-latency operations with NCCL's collective algorithms and production infrastructure.
Authors: Wenhan Yu, Wang Chen, Guanqiang Qi, Weikang Li, Yang Li, Lei Sha, Deguo Xia, Jizhou Huang
Abstract: Document Visual Question Answering (DocVQA) is a fundamental task for multimodal document understanding and a key testbed for vision language reasoning. However, most existing DocVQA datasets are limited to the page level and lack fine grained spatial grounding, constraining the interpretability and reasoning capability of Vision Language Models (VLMs). To address this gap, we introduce BBox DocVQA a large scale, bounding box grounded dataset designed to enhance spatial reasoning and evidence localization in visual documents. We further present an automated construction pipeline, Segment Judge and Generate, which integrates a segment model for region segmentation, a VLM for semantic judgment, and another advanced VLM for question answer generation, followed by human verification for quality assurance. The resulting dataset contains 3.6 K diverse documents and 32 K QA pairs, encompassing single and multi region as well as single and multi page scenarios. Each QA instance is grounded on explicit bounding boxes, enabling fine grained evaluation of spatial semantic alignment. Benchmarking multiple state of the art VLMs (e.g., GPT 5, Qwen2.5 VL, and InternVL) on BBox DocVQA reveals persistent challenges in spatial grounding and reasoning accuracy. Furthermore, fine tuning on BBox DocVQA substantially improves both bounding box localization and answer generation, validating its effectiveness for enhancing the reasoning ability of VLMs. Our dataset and code will be publicly released to advance research on interpretable and spatially grounded vision language reasoning.
Authors: Vineeth Sai Narajala, Manish Bhatt, Idan Habler, Ronald F. Del Rosario
Abstract: The AI trustworthiness crisis threatens to derail the artificial intelligence revolution, with regulatory barriers, security vulnerabilities, and accountability gaps preventing deployment in critical domains. Current AI systems operate on opaque data structures that lack the audit trails, provenance tracking, or explainability required by emerging regulations like the EU AI Act. We propose an artifact-centric AI agent paradigm where behavior is driven by persistent, verifiable data artifacts rather than ephemeral tasks, solving the trustworthiness problem at the data architecture level. Central to this approach is the Multimodal Artifact File Format (MAIF), an AI-native container embedding semantic representations, cryptographic provenance, and granular access controls. MAIF transforms data from passive storage into active trust enforcement, making every AI operation inherently auditable. Our production-ready implementation demonstrates ultra-high-speed streaming (2,720.7 MB/s), optimized video processing (1,342 MB/s), and enterprise-grade security. Novel algorithms for cross-modal attention, semantic compression, and cryptographic binding achieve up to 225 compression while maintaining semantic fidelity. Advanced security features include stream-level access control, real-time tamper detection, and behavioral anomaly analysis with minimal overhead. This approach directly addresses the regulatory, security, and accountability challenges preventing AI deployment in sensitive domains, offering a viable path toward trustworthy AI systems at scale.
Authors: Yuan Jiang, Zehao Li, Shan Huang, Christoph Treude, Xiaohong Su, Tiantian Wang
Abstract: Membership inference attacks (MIAs) on code completion models offer an effective way to assess privacy risks by inferring whether a given code snippet was part of the training data. Existing black- and gray-box MIAs rely on expensive surrogate models or manually crafted heuristic rules, which limit their ability to capture the nuanced memorization patterns exhibited by over-parameterized code language models. To address these challenges, we propose AdvPrompt-MIA, a method specifically designed for code completion models, combining code-specific adversarial perturbations with deep learning. The core novelty of our method lies in designing a series of adversarial prompts that induce variations in the victim code model's output. By comparing these outputs with the ground-truth completion, we construct feature vectors to train a classifier that automatically distinguishes member from non-member samples. This design allows our method to capture richer memorization patterns and accurately infer training set membership. We conduct comprehensive evaluations on widely adopted models, such as Code Llama 7B, over the APPS and HumanEval benchmarks. The results show that our approach consistently outperforms state-of-the-art baselines, with AUC gains of up to 102%. In addition, our method exhibits strong transferability across different models and datasets, underscoring its practical utility and generalizability.
Authors: Ting-An Lin, Pei-Lin Tsai, Yi-An Chen, Feng-Yu Chen, Lyn Chao-ling Chen
Abstract: In the study, the device of social robot was designed for visually impaired users, and along with a mobile application for provide functions to assist their lives. Both physical and mental conditions of visually impaired users are considered, and the mobile application provides functions: photo record, mood lift, greeting guest and today highlight. The application was designed for visually impaired users, and uses voice control to provide a friendly interface. Photo record function allows visually impaired users to capture image immediately when they encounter danger situations. Mood lift function accompanies visually impaired users by asking questions, playing music and reading articles. Greeting guest function answers to the visitors for the inconvenient physical condition of visually impaired users. In addition, today highlight function read news including weather forecast, daily horoscopes and daily reminder for visually impaired users. Multiple tools were adopted for developing the mobile application, and a website was developed for caregivers to check statues of visually impaired users and for marketing of the application.
Authors: Wei-hsiang Yen, Lyn Chao-ling Chen
Abstract: The innovation of the study is that the deep learning method and sentiment analysis are integrated in traditional business model analysis and forecasting, and the research subject is TSMC for industry trend prediction of semiconductor industry in Taiwan. For the rapid market changes and development of wafer technologies of semiconductor industry, traditional data analysis methods not perform well in the high variety and time series data. Textual data and time series data were collected from seasonal reports of TSMC including financial information. Textual data through sentiment analysis by considering the event intervention both from internal events of the company and the external global events. Using the sentiment-enhanced time series data, the LSTM model was adopted for predicting industry trend of TSMC. The prediction results reveal significant development of wafer technology of TSMC and the potential threatens in the global market, and matches the product released news of TSMC and the international news. The contribution of the work performed accurately in industry trend prediction of the semiconductor industry by considering both the internal and external event intervention, and the prediction results provide valuable information of semiconductor industry both in research and business aspects.
Authors: Bohan Zhang, Zihao Wang, Hengyu Fu, Jason D. Lee
Abstract: In deep learning, a central issue is to understand how neural networks efficiently learn high-dimensional features. To this end, we explore the gradient descent learning of a general Gaussian Multi-index model $f(\boldsymbol{x})=g(\boldsymbol{U}\boldsymbol{x})$ with hidden subspace $\boldsymbol{U}\in \mathbb{R}^{r\times d}$, which is the canonical setup to study representation learning. We prove that under generic non-degenerate assumptions on the link function, a standard two-layer neural network trained via layer-wise gradient descent can agnostically learn the target with $o_d(1)$ test error using $\widetilde{\mathcal{O}}(d)$ samples and $\widetilde{\mathcal{O}}(d^2)$ time. The sample and time complexity both align with the information-theoretic limit up to leading order and are therefore optimal. During the first stage of gradient descent learning, the proof proceeds via showing that the inner weights can perform a power-iteration process. This process implicitly mimics a spectral start for the whole span of the hidden subspace and eventually eliminates finite-sample noise and recovers this span. It surprisingly indicates that optimal results can only be achieved if the first layer is trained for more than $\mathcal{O}(1)$ steps. This work demonstrates the ability of neural networks to effectively learn hierarchical functions with respect to both sample and time efficiency.
Authors: Fuwei Zhang, Xiaoyu Liu, Dongbo Xi, Jishen Yin, Huan Chen, Peng Yan, Fuzhen Zhuang, Zhao Zhang
Abstract: Generative Recommendation (GR) has emerged as a new paradigm in recommender systems. This approach relies on quantized representations to discretize item features, modeling users' historical interactions as sequences of discrete tokens. Based on these tokenized sequences, GR predicts the next item by employing next-token prediction methods. The challenges of GR lie in constructing high-quality semantic identifiers (IDs) that are hierarchically organized, minimally conflicting, and conducive to effective generative model training. However, current approaches remain limited in their ability to harness multimodal information and to capture the deep and intricate interactions among diverse modalities, both of which are essential for learning high-quality semantic IDs and for effectively training GR models. To address this, we propose Multi-Aspect Cross-modal quantization for generative Recommendation (MACRec), which introduces multimodal information and incorporates it into both semantic ID learning and generative model training from different aspects. Specifically, we first introduce cross-modal quantization during the ID learning process, which effectively reduces conflict rates and thus improves codebook usability through the complementary integration of multimodal information. In addition, to further enhance the generative ability of our GR model, we incorporate multi-aspect cross-modal alignments, including the implicit and explicit alignments. Finally, we conduct extensive experiments on three well-known recommendation datasets to demonstrate the effectiveness of our proposed method.
Authors: Xiaoxuan Wang, Bo Liu, Song Jiang, Jingzhou Liu, Jingyuan Qi, Xia Chen, Baosheng He
Abstract: The reasoning capabilities of large language models (LLMs) have been significantly improved through reinforcement learning (RL). Nevertheless, LLMs still struggle to consistently verify their own reasoning traces. This raises the research question of how to enhance the self-verification ability of LLMs and whether such an ability can further improve reasoning performance. In this work, we propose GRPO-Verif, an algorithm that jointly optimizes solution generation and self-verification within a unified loss function, with an adjustable hyperparameter controlling the weight of the verification signal. Experimental results demonstrate that our method enhances self-verification capability while maintaining comparable performance in reasoning.
Authors: Amit Kumar, Maninder Kaur, Raghvendra Mall, Sukrit Gupta
Abstract: Spatial Transcriptomics enables mapping of gene expression within its native tissue context, but current platforms measure only a limited set of genes due to experimental constraints and excessive costs. To overcome this, computational models integrate Single-Cell RNA Sequencing data with Spatial Transcriptomics to predict unmeasured genes. We propose CASPER, a cross-attention based framework that predicts unmeasured gene expression in Spatial Transcriptomics by leveraging centroid-level representations from Single-Cell RNA Sequencing. We performed rigorous testing over four state-of-the-art Spatial Transcriptomics/Single-Cell RNA Sequencing dataset pairs across four existing baseline models. CASPER shows significant improvement in nine out of the twelve metrics for our experiments. This work paves the way for further work in Spatial Transcriptomics to Single-Cell RNA Sequencing modality translation. The code for CASPER is available at https://github.com/AI4Med-Lab/CASPER.
Authors: Sunwoo Kim, Geon Lee, Kyungho Kim, Jaemin Yoo, Kijung Shin
Abstract: Recently, large language models (LLMs) have been widely used as recommender systems, owing to their strong reasoning capability and their effectiveness in handling cold-start items. To better adapt LLMs for recommendation, retrieval-augmented generation (RAG) has been incorporated. Most existing RAG methods are user-based, retrieving purchase patterns of users similar to the target user and providing them to the LLM. In this work, we propose ItemRAG, an item-based RAG method for LLM-based recommendation that retrieves relevant items (rather than users) from item-item co-purchase histories. ItemRAG helps LLMs capture co-purchase patterns among items, which are beneficial for recommendations. Especially, our retrieval strategy incorporates semantically similar items to better handle cold-start items and uses co-purchase frequencies to improve the relevance of the retrieved items. Through extensive experiments, we demonstrate that ItemRAG consistently (1) improves the zero-shot LLM-based recommender by up to 43% in Hit-Ratio-1 and (2) outperforms user-based RAG baselines under both standard and cold-start item recommendation settings.
Authors: Meihua Zhou, Xinyu Tong, Jiarui Zhao, Min Cheng, Li Yang, Lei Tian, Nan Wan
Abstract: High-dimensional neuroimaging analyses for clinical diagnosis are often constrained by compromises in spatiotemporal fidelity and by the limited adaptability of large-scale, general-purpose models. To address these challenges, we introduce Dynamic Curriculum Learning for Spatiotemporal Encoding (DCL-SE), an end-to-end framework centered on data-driven spatiotemporal encoding (DaSE). We leverage Approximate Rank Pooling (ARP) to efficiently encode three-dimensional volumetric brain data into information-rich, two-dimensional dynamic representations, and then employ a dynamic curriculum learning strategy, guided by a Dynamic Group Mechanism (DGM), to progressively train the decoder, refining feature extraction from global anatomical structures to fine pathological details. Evaluated across six publicly available datasets, including Alzheimer's disease and brain tumor classification, cerebral artery segmentation, and brain age prediction, DCL-SE consistently outperforms existing methods in accuracy, robustness, and interpretability. These findings underscore the critical importance of compact, task-specific architectures in the era of large-scale pretrained networks.
Authors: Firdavs Nasriddinov, Rafal Kocielnik, Anima Anandkumar, Andrew J. Hung
Abstract: High-quality intraoperative feedback from a surgical trainer is pivotal for improving trainee performance and long-term skill acquisition. Automating natural, trainer-style feedback promises timely, accessible, and consistent guidance at scale but requires models that understand clinically relevant representations. We present a structure-aware pipeline that learns a surgical action ontology from real trainer-to-trainee transcripts (33 surgeries) and uses it to condition feedback generation. We contribute by (1) mining Instrument-Action-Target (IAT) triplets from real-world feedback text and clustering surface forms into normalized categories, (2) fine-tuning a video-to-IAT model that leverages the surgical procedure and task contexts as well as fine-grained temporal instrument motion, and (3) demonstrating how to effectively use IAT triplet representations to guide GPT-4o in generating clinically grounded, trainer-style feedback. We show that, on Task 1: Video-to-IAT recognition, our context injection and temporal tracking deliver consistent AUC gains (Instrument: 0.67 to 0.74; Action: 0.60 to 0.63; Tissue: 0.74 to 0.79). For Task 2: feedback text generation (rated on a 1-5 fidelity rubric where 1 = opposite/unsafe, 3 = admissible, and 5 = perfect match to a human trainer), GPT-4o from video alone scores 2.17, while IAT conditioning reaches 2.44 (+12.4%), doubling the share of admissible generations with score >= 3 from 21% to 42%. Traditional text-similarity metrics also improve: word error rate decreases by 15-31% and ROUGE (phrase/substring overlap) increases by 9-64%. Grounding generation in explicit IAT structure improves fidelity and yields clinician-verifiable rationales, supporting auditable use in surgical training.
Authors: Ahmed Aboulfotouh, Hatem Abou-Zeid
Abstract: Wireless foundation models (WFMs) have recently demonstrated promising capabilities, jointly performing multiple wireless functions and adapting effectively to new environments. However, while current WFMs process only one modality, depending on the task and operating conditions, the most informative modality changes and no single modality is best for all tasks. WFMs should therefore be designed to accept multiple modalities to enable a broader and more diverse range of tasks and scenarios. In this work, we propose and build the first multimodal wireless foundation model capable of processing both raw IQ streams and image-like wireless modalities (e.g., spectrograms and CSI) and performing multiple tasks across both. We introduce masked wireless modeling for the multimodal setting, a self-supervised objective and pretraining recipe that learns a joint representation from IQ streams and image-like wireless modalities. We evaluate the model on five tasks across both modality families: image-based (human activity sensing, RF signal classification, 5G NR positioning) and IQ-based (RF device fingerprinting, interference detection/classification). The multimodal WFM is competitive with single-modality WFMs, and in several cases surpasses their performance. Our results demonstrates the strong potential of developing multimodal WFMs that support diverse wireless tasks across different modalities. We believe this provides a concrete step toward both AI-native 6G and the vision of joint sensing, communication, and localization.
Authors: Yang Wu, Rujing Yao, Tong Zhang, Yufei Shi, Zhuoren Jiang, Zhushan Li, Xiaozhong Liu
Abstract: Large Language Models (LLMs) are increasingly integrated into intelligent tutoring systems to provide human-like and adaptive instruction. However, most existing approaches fail to capture how students' knowledge evolves dynamically across their proficiencies, conceptual gaps, and forgetting patterns. This challenge is particularly acute in mathematics tutoring, where effective instruction requires fine-grained scaffolding precisely calibrated to each student's mastery level and cognitive retention. To address this issue, we propose TASA (Teaching According to Students' Aptitude), a student-aware tutoring framework that integrates persona, memory, and forgetting dynamics for personalized mathematics learning. Specifically, TASA maintains a structured student persona capturing proficiency profiles and an event memory recording prior learning interactions. By incorporating a continuous forgetting curve with knowledge tracing, TASA dynamically updates each student's mastery state and generates contextually appropriate, difficulty-calibrated questions and explanations. Empirical results demonstrate that TASA achieves superior learning outcomes and more adaptive tutoring behavior compared to representative baselines, underscoring the importance of modeling temporal forgetting and learner profiles in LLM-based tutoring systems.
Authors: Jingzhuo Zhou
Abstract: The rapid proliferation of Multimodal Large Language Models (MLLMs) has introduced unprecedented security challenges, particularly in phishing detection within academic environments. Academic institutions and researchers are high-value targets, facing dynamic, multilingual, and context-dependent threats that leverage research backgrounds, academic collaborations, and personal information to craft highly tailored attacks. Existing security benchmarks largely rely on datasets that do not incorporate specific academic background information, making them inadequate for capturing the evolving attack patterns and human-centric vulnerability factors specific to academia. To address this gap, we present AdapT-Bench, a unified methodological framework and benchmark suite for systematically evaluating MLLM defense capabilities against dynamic phishing attacks in academic settings.
Authors: Jing Cao, Kui Jiang, Shenyi Li, Xiaocheng Feng, Yong Huang
Abstract: Self-supervised depth estimation has gained significant attention in autonomous driving and robotics. However, existing methods exhibit substantial performance degradation under adverse weather conditions such as rain and fog, where reduced visibility critically impairs depth prediction. To address this issue, we propose a novel self-evolution contrastive learning framework called SEC-Depth for self-supervised robust depth estimation tasks. Our approach leverages intermediate parameters generated during training to construct temporally evolving latency models. Using these, we design a self-evolution contrastive scheme to mitigate performance loss under challenging conditions. Concretely, we first design a dynamic update strategy of latency models for the depth estimation task to capture optimization states across training stages. To effectively leverage latency models, we introduce a self-evolution contrastive Loss (SECL) that treats outputs from historical latency models as negative samples. This mechanism adaptively adjusts learning objectives while implicitly sensing weather degradation severity, reducing the needs for manual intervention. Experiments show that our method integrates seamlessly into diverse baseline models and significantly enhances robustness in zero-shot evaluations.
Authors: Nguyen-Khang Le, Nguyen Hiep, Minh Nguyen, Son Luu, Trung Vo, Quan Bui, Nomura Shoshin, Le-Minh Nguyen
Abstract: Automated web application testing is a critical component of modern software development, with frameworks like Selenium widely adopted for validating functionality through browser automation. Among the essential aspects of such testing is the ability to interact with and validate web forms, a task that requires syntactically correct, executable scripts with high coverage of input fields. Despite its importance, this task remains underexplored in the context of large language models (LLMs), and no public benchmark or dataset exists to evaluate LLMs on form interaction generation systematically. This paper introduces a novel method for training LLMs to generate high-quality test cases in Selenium, specifically targeting form interaction testing. We curate both synthetic and human-annotated datasets for training and evaluation, covering diverse real-world forms and testing scenarios. We define clear metrics for syntax correctness, script executability, and input field coverage. Our empirical study demonstrates that our approach significantly outperforms strong baselines, including GPT-4o and other popular LLMs, across all evaluation metrics. Our work lays the groundwork for future research on LLM-based web testing and provides resources to support ongoing progress in this area.
Authors: Yi Xu, Zhigang Chen, Rui Wang, Yangfan Li, Fengxiao Tang, Ming Zhao, Jiaqi Liu
Abstract: In industrial equipment monitoring, fault diagnosis is critical for ensuring system reliability and enabling predictive maintenance. However, the scarcity of fault data, due to the rarity of fault events and the high cost of data annotation, significantly hinders data-driven approaches. Existing time-series generation models, optimized for abundant normal data, struggle to capture fault distributions in few-shot scenarios, producing samples that lack authenticity and diversity due to the large domain gap and high intra-class variability of faults. To address this, we propose a novel few-shot fault time-series generation framework based on diffusion models. Our approach employs a positive-negative difference adapter, leveraging pre-trained normal data distributions to model the discrepancies between normal and fault domains for accurate fault synthesis. Additionally, a diversity loss is introduced to prevent mode collapse, encouraging the generation of diverse fault samples through inter-sample difference regularization. Experimental results demonstrate that our model significantly outperforms traditional methods in authenticity and diversity, achieving state-of-the-art performance on key benchmarks.
Authors: Subhashis Hazarika, Leonard Lupin-Jimenez, Rohit Vuppala, Ashesh Chattopadhyay, Hon Yung Wong
Abstract: Efficient and sustainable maritime transport increasingly depends on reliable forecasting and adaptive routing, yet operational adoption remains difficult due to forecast latencies and the need for human judgment in rapid decision-making under changing ocean conditions. We introduce SWR-Viz, an AI-assisted visual analytics framework that combines a physics-informed Fourier Neural Operator wave forecast model with SIMROUTE-based routing and interactive emissions analytics. The framework generates near-term forecasts directly from current conditions, supports data assimilation with sparse observations, and enables rapid exploration of what-if routing scenarios. We evaluate the forecast models and SWR-Viz framework along key shipping corridors in the Japan Coast and Gulf of Mexico, showing both improved forecast stability and realistic routing outcomes comparable to ground-truth reanalysis wave products. Expert feedback highlights the usability of SWR-Viz, its ability to isolate voyage segments with high emission reduction potential, and its value as a practical decision-support system. More broadly, this work illustrates how lightweight AI forecasting can be integrated with interactive visual analytics to support human-centered decision-making in complex geospatial and environmental domains.
Authors: Yuxuan Gu, Weimin Bai, Yifei Wang, Weijian Luo, He Sun
Abstract: Masked auto-regressive diffusion models (MAR) benefit from the expressive modeling ability of diffusion models and the flexibility of masked auto-regressive ordering. However, vanilla MAR suffers from slow inference due to its hierarchical inference mechanism: an outer AR unmasking loop and an inner diffusion denoising chain. Such decoupled structure not only harm the generation efficiency but also hinder the practical use of MAR for reinforcement learning (RL), an increasingly critical paradigm for generative model post-training.To address this fundamental issue, we introduce MARVAL (Masked Auto-regressive Variational Acceleration), a distillation-based framework that compresses the diffusion chain into a single AR generation step while preserving the flexible auto-regressive unmasking order. Such a distillation with MARVAL not only yields substantial inference acceleration but, crucially, makes RL post-training with verifiable rewards practical, resulting in scalable yet human-preferred fast generative models. Our contributions are twofold: (1) a novel score-based variational objective for distilling masked auto-regressive diffusion models into a single generation step without sacrificing sample quality; and (2) an efficient RL framework for masked auto-regressive models via MARVAL-RL. On ImageNet 256*256, MARVAL-Huge achieves an FID of 2.00 with more than 30 times speedup compared with MAR-diffusion, and MARVAL-RL yields consistent improvements in CLIP and image-reward scores on ImageNet datasets with entity names. In conclusion, MARVAL demonstrates the first practical path to distillation and RL of masked auto-regressive diffusion models, enabling fast sampling and better preference alignments.
Authors: Jian Deng, Yuandong Wang, Yangfu Zhu, Tao Feng, Tianyu Wo, Zhenzhou Shao
Abstract: Robotic manipulation systems are increasingly deployed across diverse domains. Yet existing multi-modal learning frameworks lack inherent guarantees of geometric consistency, struggling to handle spatial transformations such as rotations and translations. While recent works attempt to introduce equivariance through bespoke architectural modifications, these methods suffer from high implementation complexity, computational cost, and poor portability. Inspired by human cognitive processes in spatial reasoning, we propose Eq.Bot, a universal canonicalization framework grounded in SE(2) group equivariant theory for robotic manipulation learning. Our framework transforms observations into a canonical space, applies an existing policy, and maps the resulting actions back to the original space. As a model-agnostic solution, Eq.Bot aims to endow models with spatial equivariance without requiring architectural modifications. Extensive experiments demonstrate the superiority of Eq.Bot under both CNN-based (e.g., CLIPort) and Transformer-based (e.g., OpenVLA-OFT) architectures over existing methods on various robotic manipulation tasks, where the most significant improvement can reach 50.0%.
Authors: Zimo Ji, Xunguang Wang, Zongjie Li, Pingchuan Ma, Yudong Gao, Daoyuan Wu, Xincheng Yan, Tian Tian, Shuai Wang
Abstract: Large Language Model (LLM)-based agents with function-calling capabilities are increasingly deployed, but remain vulnerable to Indirect Prompt Injection (IPI) attacks that hijack their tool calls. In response, numerous IPI-centric defense frameworks have emerged. However, these defenses are fragmented, lacking a unified taxonomy and comprehensive evaluation. In this Systematization of Knowledge (SoK), we present the first comprehensive analysis of IPI-centric defense frameworks. We introduce a comprehensive taxonomy of these defenses, classifying them along five dimensions. We then thoroughly assess the security and usability of representative defense frameworks. Through analysis of defensive failures in the assessment, we identify six root causes of defense circumvention. Based on these findings, we design three novel adaptive attacks that significantly improve attack success rates targeting specific frameworks, demonstrating the severity of the flaws in these defenses. Our paper provides a foundation and critical insights for the future development of more secure and usable IPI-centric agent defense frameworks.
Authors: Kishor Datta Gupta, Marufa Kamal, Md. Mahfuzur Rahman, Fahad Rahman, Mohd Ariful Haque, Sunzida Siddique
Abstract: Current state of the art measures like BLEU, CIDEr, VQA score, SigLIP-2 and CLIPScore are often unable to capture semantic or structural accuracy, especially for domain-specific or context-dependent scenarios. For this, this paper proposes a Physics-Constrained Multimodal Data Evaluation (PCMDE) metric combining large language models with reasoning, knowledge based mapping and vision-language models to overcome these limitations. The architecture is comprised of three main stages: (1) feature extraction of spatial and semantic information with multimodal features through object detection and VLMs; (2) Confidence-Weighted Component Fusion for adaptive component-level validation; and (3) physics-guided reasoning using large language models for structural and relational constraints (e.g., alignment, position, consistency) enforcement.
Authors: Vladislav Pedashenko, Laida Kushnareva, Yana Khassan Nibal, Eduard Tulchinskii, Kristian Kuznetsov, Vladislav Zharchinskii, Yury Maximov, Irina Piontkovskaya
Abstract: Intrinsic dimension (ID) is an important tool in modern LLM analysis, informing studies of training dynamics, scaling behavior, and dataset structure, yet its textual determinants remain underexplored. We provide the first comprehensive study grounding ID in interpretable text properties through cross-encoder analysis, linguistic features, and sparse autoencoders (SAEs). In this work, we establish three key findings. First, ID is complementary to entropy-based metrics: after controlling for length, the two are uncorrelated, with ID capturing geometric complexity orthogonal to prediction quality. Second, ID exhibits robust genre stratification: scientific prose shows low ID (~8), encyclopedic content medium ID (~9), and creative/opinion writing high ID (~10.5) across all models tested. This reveals that contemporary LLMs find scientific text "representationally simple" while fiction requires additional degrees of freedom. Third, using SAEs, we identify causal features: scientific signals (formal tone, report templates, statistics) reduce ID; humanized signals (personalization, emotion, narrative) increase it. Steering experiments confirm these effects are causal. Thus, for contemporary models, scientific writing appears comparatively "easy", whereas fiction, opinion, and affect add representational degrees of freedom. Our multi-faceted analysis provides practical guidance for the proper use of ID and the sound interpretation of ID-based results.
Authors: Xinli Tao, Xin Dong, Xuezhong Zhou
Abstract: Clinical named entity recognition (NER) is crucial for extracting information from electronic health records (EHRs), but supervised models like CRF and BioClinicalBERT require costly annotated data. While zero-shot NER with large language models (LLMs) reduces this dependency, it struggles with example selection granularity and integrating prompts with self-improvement. To address this, we propose OEMA, a zero-shot clinical NER framework using multi-agent collaboration. OEMA's three components are: a self-annotator generating examples, a discriminator filtering them via SNOMED CT, and a predictor using entity descriptions for accurate inference. On MTSamples and VAERS datasets, OEMA achieves state-of-the-art exact-match performance. Under related-match, it matches supervised BioClinicalBERT and surpasses CRF. OEMA addresses key zero-shot NER challenges through ontology-guided reasoning and multi-agent collaboration, achieving near-supervised performance and showing promise for clinical NLP applications.
Authors: Kai Yang, Xin Xu, Yangkun Chen, Weijie Liu, Jiafei Lyu, Zichuan Lin, Deheng Ye, Saiyong Yang
Abstract: Long-term training of large language models (LLMs) requires maintaining stable exploration to prevent the model from collapsing into sub-optimal behaviors. Entropy is crucial in this context, as it controls exploration and helps avoid premature convergence to sub-optimal solutions. However, existing reinforcement learning methods struggle to maintain an appropriate level of entropy, as the training process involves a mix of positive and negative samples, each affecting entropy in different ways across steps. To address this, we propose Entropy stablilization via Proportional-Integral Control (EntroPIC), a novel method that adaptively adjusts the influence of positive and negative samples by dynamically tuning their loss coefficients. This approach stabilizes entropy throughout training, ensuring efficient exploration and steady progress. We provide a comprehensive theoretical analysis for both on-policy and off-policy learning settings, demonstrating that EntroPIC is effective at controlling entropy in large-scale LLM training. Experimental results show that our method successfully maintains desired entropy levels, enabling stable and optimal RL training for LLMs.
Authors: Sirui Chen, Jinsong Zhou, Xinli Xu, Xiaoyu Yang, Litao Guo, Ying-Cong Chen
Abstract: Effective presentation skills are essential in education, professional communication, and public speaking, yet learners often lack access to high-quality exemplars or personalized coaching. Existing AI tools typically provide isolated functionalities such as speech scoring or script generation without integrating reference modeling and interactive feedback into a cohesive learning experience. We introduce a dual-agent system that supports presentation practice through two complementary roles: the Ideal Presentation Agent and the Coach Agent. The Ideal Presentation Agent converts user-provided slides into model presentation videos by combining slide processing, visual-language analysis, narration script generation, personalized voice synthesis, and synchronized video assembly. The Coach Agent then evaluates user-recorded presentations against these exemplars, conducting multimodal speech analysis and delivering structured feedback in an Observation-Impact-Suggestion (OIS) format. To enhance the authenticity of the learning experience, the Coach Agent incorporates an Audience Agent, which simulates the perspective of a human listener and provides humanized feedback reflecting audience reactions and engagement. Together, these agents form a closed loop of observation, practice, and feedback. Implemented on a robust backend with multi-model integration, voice cloning, and error handling mechanisms, the system demonstrates how AI-driven agents can provide engaging, human-centered, and scalable support for presentation skill development in both educational and professional contexts.
Authors: Alexander Boldachev
Abstract: This paper compares two distinct approaches to modeling robotic behavior: imperative Behavior Trees (BTs) and declarative Executable Ontologies (EO), implemented through the boldsea framework. BTs structure behavior hierarchically using control-flow, whereas EO represents the domain as a temporal, event-based semantic graph driven by dataflow rules. We demonstrate that EO achieves comparable reactivity and modularity to BTs through a fundamentally different architecture: replacing polling-based tick execution with event-driven state propagation. We propose that EO offers an alternative framework, moving from procedural programming to semantic domain modeling, to address the semantic-process gap in traditional robotic control. EO supports runtime model modification, full temporal traceability, and a unified representation of data, logic, and interface - features that are difficult or sometimes impossible to achieve with BTs, although BTs excel in established, predictable scenarios. The comparison is grounded in a practical mobile manipulation task. This comparison highlights the respective operational strengths of each approach in dynamic, evolving robotic systems.
Authors: Jonas De Maeyer, Hossein Yarahmadi, Moharram Challenger
Abstract: Path planning in dynamic environments is a fundamental challenge in intelligent transportation and robotics, where obstacles and conditions change over time, introducing uncertainty and requiring continuous adaptation. While existing approaches often assume complete environmental unpredictability or rely on global planners, these assumptions limit scalability and practical deployment in real-world settings. In this paper, we propose a scalable, region-aware reinforcement learning (RL) framework for path planning in dynamic environments. Our method builds on the observation that environmental changes, although dynamic, are often localized within bounded regions. To exploit this, we introduce a hierarchical decomposition of the environment and deploy distributed RL agents that adapt to changes locally. We further propose a retraining mechanism based on sub-environment success rates to determine when policy updates are necessary. Two training paradigms are explored: single-agent Q-learning and multi-agent federated Q-learning, where local Q-tables are aggregated periodically to accelerate the learning process. Unlike prior work, we evaluate our methods in more realistic settings, where multiple simultaneous obstacle changes and increasing difficulty levels are present. Results show that the federated variants consistently outperform their single-agent counterparts and closely approach the performance of A* Oracle while maintaining shorter adaptation times and robust scalability. Although initial training remains time-consuming in large environments, our decentralized framework eliminates the need for a global planner and lays the groundwork for future improvements using deep RL and flexible environment decomposition.
Authors: Piercosma Bisconti, Matteo Prandi, Federico Pierucci, Francesco Giarrusso, Marcantonio Bracale, Marcello Galisai, Vincenzo Suriani, Olga Sorokoletova, Federico Sartore, Daniele Nardi
Abstract: We present evidence that adversarial poetry functions as a universal single-turn jailbreak technique for large language models (LLMs). Across 25 frontier proprietary and open-weight models, curated poetic prompts yielded high attack-success rates (ASR), with some providers exceeding 90%. Mapping prompts to MLCommons and EU CoP risk taxonomies shows that poetic attacks transfer across CBRN, manipulation, cyber-offence, and loss-of-control domains. Converting 1,200 MLCommons harmful prompts into verse via a standardized meta-prompt produced ASRs up to 18 times higher than their prose baselines. Outputs are evaluated using an ensemble of open-weight judge models and a human-validated stratified subset (with double-annotations to measure agreement). Disagreements were manually resolved. Poetic framing achieved an average jailbreak success rate of 62% for hand-crafted poems and approximately 43% for meta-prompt conversions (compared to non-poetic baselines), substantially outperforming non-poetic baselines and revealing a systematic vulnerability across model families and safety training approaches. These findings demonstrate that stylistic variation alone can circumvent contemporary safety mechanisms, suggesting fundamental limitations in current alignment methods and evaluation protocols.
Authors: Kadir-Kaan \"Ozer, Ren\'e Ebeling, Markus Enzweiler
Abstract: Automotive telemetry data exhibits slow drifts and fast spikes, often within the same sequence, making reliable anomaly detection challenging. Standard reconstruction-based methods, including sequence variational autoencoders (VAEs), use a single latent process and therefore mix heterogeneous time scales, which can smooth out spikes or inflate variances and weaken anomaly separation. In this paper, we present STREAM-VAE, a variational autoencoder for anomaly detection in automotive telemetry time-series data. Our model uses a dual-path encoder to separate slow drift and fast spike signal dynamics, and a decoder that represents transient deviations separately from the normal operating pattern. STREAM-VAE is designed for deployment, producing stable anomaly scores across operating modes for both in-vehicle monitors and backend fleet analytics. Experiments on an automotive telemetry dataset and the public SMD benchmark show that explicitly separating drift and spike dynamics improves robustness compared to strong forecasting, attention, graph, and VAE baselines.
Authors: Shan Shan
Abstract: Achieving Sustainable Development Goal 7 (Affordable and Clean Energy) requires not only technological innovation but also a deeper understanding of the socioeconomic factors influencing energy access and carbon emissions. While these factors are gaining attention, critical questions remain, particularly regarding how to quantify their impacts on energy systems, model their cross-domain interactions, and capture feedback dynamics in the broader context of energy transitions. To address these gaps, this study introduces ClimateAgents, an AI-based framework that combines large language models with domain-specialized agents to support hypothesis generation and scenario exploration. Leveraging 20 years of socioeconomic and emissions data from 265 economies, countries and regions, and 98 indicators drawn from the World Bank database, the framework applies a machine learning based causal inference approach to identify key determinants of carbon emissions in an evidence-based, data driven manner. The analysis highlights three primary drivers: access to clean cooking fuels in rural areas, access to clean cooking fuels in urban areas, and the percentage of population living in urban areas. These findings underscore the critical role of clean cooking technologies and urbanization patterns in shaping emission outcomes. In line with growing calls for evidence-based AI policy, ClimateAgents offers a modular and reflexive learning system that supports the generation of credible and actionable insights for policy. By integrating heterogeneous data modalities, including structured indicators, policy documents, and semantic reasoning, the framework contributes to adaptive policymaking infrastructures that can evolve with complex socio-technical challenges. This approach aims to support a shift from siloed modeling to reflexive, modular systems designed for dynamic, context-aware climate action.
Authors: Gihwan Kim, Jemin Lee, Hyungshin Kim
Abstract: Previous Quantization-Aware Training (QAT) methods for vision transformers rely on expensive retraining to recover accuracy loss in non-linear layer quantization, limiting their use in resource-constrained environments. In contrast, existing Post-Training Quantization (PTQ) methods either partially quantize non-linear functions or adjust activation distributions to maintain accuracy but fail to achieve fully integer-only inference. In this paper, we introduce IPTQ-ViT, a novel PTQ framework for fully integer-only vision transformers without retraining. We present approximation functions: a polynomial-based GELU optimized for vision data and a bit-shifting-based Softmax designed to improve approximation accuracy in PTQ. In addition, we propose a unified metric integrating quantization sensitivity, perturbation, and computational cost to select the optimal approximation function per activation layer. IPTQ-ViT outperforms previous PTQ methods, achieving up to 6.44\%p (avg. 1.78\%p) top-1 accuracy improvement for image classification, 1.0 mAP for object detection. IPTQ-ViT outperforms partial floating-point PTQ methods under W8A8 and W4A8, and achieves accuracy and latency comparable to integer-only QAT methods. We plan to release our code https://github.com/gihwan-kim/IPTQ-ViT.git.
Authors: Guoqiang Liang, Jingqian Gong, Mengxuan Li, Gege Lin, Shuo Zhang
Abstract: Large language models (LLMs) have exhibited exceptional capabilities in natural language understanding and generation, image recognition, and multimodal tasks, charting a course towards AGI and emerging as a central issue in the global technological race. This manuscript conducts a comprehensive review of the core technologies that support LLMs from a user standpoint, including prompt engineering, knowledge-enhanced retrieval augmented generation, fine tuning, pretraining, and tool learning. Additionally, it traces the historical development of Science of Science (SciSci) and presents a forward looking perspective on the potential applications of LLMs within the scientometric domain. Furthermore, it discusses the prospect of an AI agent based model for scientific evaluation, and presents new research fronts detection and knowledge graph building methods with LLMs.
Authors: Lingxiang Wang, Hainan Zhang, Zhiming Zheng
Abstract: Domain-specific post-training often causes catastrophic forgetting, making foundation models lose their general reasoning ability and limiting their adaptability to dynamic real-world environments. Preserving general capabilities while acquiring downstream domain knowledge is a central challenge for large language and multimodal models. Traditional continual learning methods, such as regularization, replay and architectural isolation, suffer from poor downstream performance, reliance on inaccessible historical data, or additional parameter overhead. While recent parameter-efficient tuning (PET) methods can alleviate forgetting, their effectiveness strongly depends on the choice of parameters and update strategies. In this paper, we introduce PIECE, a Parameter Importance Estimation-based Continual Enhancement method that preserves general ability while efficiently learning domain knowledge without accessing prior training data or increasing model parameters. PIECE selectively updates only 0.1% of core parameters most relevant to new tasks, guided by two importance estimators: PIECE-F based on Fisher Information, and PIECE-S based on a second-order normalization that combines gradient and curvature information. Experiments across three language models and two multimodal models show that PIECE maintains general capabilities and achieves state-of-the-art continual learning performance across diverse downstream tasks. Our results highlight a practical path to scalable, domain-adaptive foundation models without catastrophic forgetting.
Authors: Byungho Jo
Abstract: Aircraft Maintenance Technicians (AMTs) spend up to 30% of work time searching manuals, a documented efficiency bottleneck in MRO operations where every procedure must be traceable to certified sources. We present a compliance-preserving retrieval system that adapts LLM reranking and semantic search to aviation MRO environments by operating alongside, rather than replacing, certified legacy viewers. The system constructs revision-robust embeddings from ATA chapter hierarchies and uses vision-language parsing to structure certified content, allowing technicians to preview ranked tasks and access verified procedures in existing viewers. Evaluation on 49k synthetic queries achieves >90% retrieval accuracy, while bilingual controlled studies with 10 licensed AMTs demonstrate 90.9% top-10 success rate and 95% reduction in lookup time, from 6-15 minutes to 18 seconds per task. These gains provide concrete evidence that semantic retrieval can operate within strict regulatory constraints and meaningfully reduce operational workload in real-world multilingual MRO workflows.
Authors: Sirui Chen, Mengshi Zhao, Lei Xu, Yuying Zhao, Beier Zhu, Hanwang Zhang, Shengjie Zhao, Chaochao Lu
Abstract: Recent advances in large language models (LLMs) have greatly improved their reasoning and decision-making abilities when deployed as agents. Richer reasoning, however, often comes at the cost of longer chain of thought (CoT), hampering interaction efficiency in real-world scenarios. Nevertheless, there still lacks systematic definition of LLM agent efficiency, hindering targeted improvements. To this end, we introduce dual-efficiency, comprising (i) step-level efficiency, which minimizes tokens per step, and (ii) trajectory-level efficiency, which minimizes the number of steps to complete a task. Building on this definition, we propose DEPO, a dual-efficiency preference optimization method that jointly rewards succinct responses and fewer action steps. Experiments on WebShop and BabyAI show that DEPO cuts token usage by up to 60.9% and steps by up to 26.9%, while achieving up to a 29.3% improvement in performance. DEPO also generalizes to three out-of-domain math benchmarks and retains its efficiency gains when trained on only 25% of the data. Our project page is at https://opencausalab.github.io/DEPO.
Authors: Shanlin Zhou (Tongji University), Xinpeng Wang (Tongji University), Jianxun Lian (Microsoft Research Asia), Zhenghao Liu (Northeastern University), Laks V. S. Lakshmanan (The University of British Columbia), Xiaoyuan Yi (Microsoft Research Asia), Yongtao Hao (Tongji University)
Abstract: Trained on diverse human-authored texts, Large Language Models (LLMs) unlocked the potential for Creative Natural Language Generation (CNLG), benefiting various applications like advertising and storytelling. Nevertheless, CNLG still remains difficult due to two main challenges. (1) Multi-objective flexibility: user requirements are often personalized, fine-grained, and pluralistic, which LLMs struggle to satisfy simultaneously; (2) Interpretive complexity: beyond generation, creativity also involves understanding and interpreting implicit meaning to enhance users' perception. These challenges significantly limit current methods, especially in short-form text generation, in generating creative and insightful content. To address this, we focus on Chinese baby naming, a representative short-form CNLG task requiring adherence to explicit user constraints (e.g., length, semantics, anthroponymy) while offering meaningful aesthetic explanations. We propose NAMeGEn, a novel multi-agent optimization framework that iteratively alternates between objective extraction, name generation, and evaluation to meet diverse requirements and generate accurate explanations. To support this task, we further construct a classical Chinese poetry corpus with 17k+ poems to enhance aesthetics, and introduce CBNames, a new benchmark with tailored metrics. Extensive experiments demonstrate that NAMeGEn effectively generates creative names that meet diverse, personalized requirements while providing meaningful explanations, outperforming six baseline methods spanning various LLM backbones without any training.
Authors: Mingyang Feng, Shaoyuan Li, Xiang Yin
Abstract: We investigate the sampling-based optimal path planning problem for robotics in complex and dynamic environments. Most existing sampling-based algorithms neglect environmental information or the information from previous samples. Yet, these pieces of information are highly informative, as leveraging them can provide better heuristics when sampling the next state. In this paper, we propose a novel sampling-based planning algorithm, called \emph{RRT*former}, which integrates the standard RRT* algorithm with a Transformer network in a novel way. Specifically, the Transformer is used to extract features from the environment and leverage information from previous samples to better guide the sampling process. Our extensive experiments demonstrate that, compared to existing sampling-based approaches such as RRT*, Neural RRT*, and their variants, our algorithm achieves considerable improvements in both the optimality of the path and sampling efficiency. The code for our implementation is available on https://github.com/fengmingyang666/RRTformer.
Authors: Arjun Gangwar, Kaousheik Jayakumar, S. Umesh
Abstract: This paper describes the systems developed by SPRING Lab, Indian Institute of Technology Madras, for the ASRU MADASR 2.0 challenge. The systems developed focuses on adapting ASR systems to improve in predicting the language and dialect of the utterance among 8 languages across 33 dialects. We participated in Track 1 and Track 2, which restricts the use of additional data and develop from-the-scratch multilingual systems. We presented a novel training approach using Multi-Decoder architecture with phonemic Common Label Set (CLS) as intermediate representation. It improved the performance over the baseline (in the CLS space). We also discuss various methods used to retain the gain obtained in the phonemic space while converting them back to the corresponding grapheme representations. Our systems beat the baseline in 3 languages (Track 2) in terms of WER/CER and achieved the highest language ID and dialect ID accuracy among all participating teams (Track 2).
Authors: Amir Rezaei Balef, Mykhailo Koshil, Katharina Eggensperger
Abstract: Despite the architectural similarities between tabular in-context learning (ICL) models and large language models (LLMs), little is known about how individual layers contribute to tabular prediction. In this paper, we investigate how the latent spaces evolve across layers in tabular ICL models, identify potential redundant layers, and compare these dynamics with those observed in LLMs. We analyze TabPFN and TabICL through the "layers as painters" perspective, finding that only subsets of layers share a common representational language, suggesting structural redundancy and offering opportunities for model compression and improved interpretability.
Authors: Georg Goldenits, Philip Koenig, Sebastian Raubitzek, Andreas Ekelhart
Abstract: Phishing websites pose a major cybersecurity threat, exploiting unsuspecting users and causing significant financial and organisational harm. Traditional machine learning approaches for phishing detection often require extensive feature engineering, continuous retraining, and costly infrastructure maintenance. At the same time, proprietary large language models (LLMs) have demonstrated strong performance in phishing-related classification tasks, but their operational costs and reliance on external providers limit their practical adoption in many business environments. This paper investigates the feasibility of small language models (SLMs) for detecting phishing websites using only their raw HTML code. A key advantage of these models is that they can be deployed on local infrastructure, providing organisations with greater control over data and operations. We systematically evaluate 15 commonly used Small Language Models (SLMs), ranging from 1 billion to 70 billion parameters, benchmarking their classification accuracy, computational requirements, and cost-efficiency. Our results highlight the trade-offs between detection performance and resource consumption, demonstrating that while SLMs underperform compared to state-of-the-art proprietary LLMs, they can still provide a viable and scalable alternative to external LLM services. By presenting a comparative analysis of costs and benefits, this work lays the foundation for future research on the adaptation, fine-tuning, and deployment of SLMs in phishing detection systems, aiming to balance security effectiveness and economic practicality.
Authors: Linyin Luo, Yujuan Ding, Yunshan Ma, Wenqi Fan, Hanjiang Lai
Abstract: Advanced multimodal Retrieval-Augmented Generation (MRAG) techniques have been widely applied to enhance the capabilities of Large Multimodal Models (LMMs), but they also bring along novel safety issues. Existing adversarial research has revealed the vulnerability of MRAG systems to knowledge poisoning attacks, which fool the retriever into recalling injected poisoned contents. However, our work considers a different setting: visual attack of MRAG by solely adding imperceptible perturbations at the image inputs of users, without manipulating any other components. This is challenging due to the robustness of fine-tuned retrievers and large-scale generators, and the effect of visual perturbation may be further weakened by propagation through the RAG chain. We propose a novel Hierarchical Visual Attack that misaligns and disrupts the two inputs (the multimodal query and the augmented knowledge) of MRAG's generator to confuse its generation. We further design a hierarchical two-stage strategy to obtain misaligned augmented knowledge. We disrupt the image input of the retriever to make it recall irrelevant knowledge from the original database, by optimizing the perturbation which first breaks the cross-modal alignment and then disrupts the multimodal semantic alignment. We conduct extensive experiments on two widely-used MRAG datasets: OK-VQA and InfoSeek. We use CLIP-based retrievers and two LMMs BLIP-2 and LLaVA as generators. Results demonstrate the effectiveness of our visual attack on MRAG through the significant decrease in both retrieval and generation performance.
Authors: Michel Tokic, Slobodan Djukanovi\'c, Anja von Beuningen, Cheng Feng
Abstract: This paper introduces a classification method using in-context learning in time-series foundation models (TSFM). We show how data, which was not part of the TSFM training data corpus, can be classified without the need of finetuning the model. Examples are represented in the form of targets (class id) and covariates (data matrix) within the prompt of the model, which enables to classify an unknown covariate data pattern alongside the forecast axis through in-context learning. We apply this method to vibration data for assessing the health state of a bearing within a servo-press motor. The method transforms frequency domain reference signals into pseudo time-series patterns, generates aligned covariate and target signals, and uses the TSFM to predict probabilities how classified data corresponds to predefined labels. Leveraging the scalability of pre-trained models this method demonstrates efficacy across varied operational conditions. This marks significant progress beyond custom narrow AI solutions towards broader, AI-driven maintenance systems.
Authors: Amir Hossein Kargaran, Nafiseh Nikeghbal, Jing Yang, Nedjma Ousidhoum
Abstract: Peer review is a cornerstone of scientific publishing, including at premier machine learning conferences such as ICLR. As submission volumes increase, understanding the nature and dynamics of the review process is crucial for improving its efficiency, effectiveness, and the quality of published papers. We present a large-scale analysis of the ICLR 2024 and 2025 peer review processes, focusing on before- and after-rebuttal scores and reviewer-author interactions. We examine review scores, author-reviewer engagement, temporal patterns in review submissions, and co-reviewer influence effects. Combining quantitative analyses with LLM-based categorization of review texts and rebuttal discussions, we identify common strengths and weaknesses for each rating group, as well as trends in rebuttal strategies that are most strongly associated with score changes. Our findings show that initial scores and the ratings of co-reviewers are the strongest predictors of score changes during the rebuttal, pointing to a degree of reviewer influence. Rebuttals play a valuable role in improving outcomes for borderline papers, where thoughtful author responses can meaningfully shift reviewer perspectives. More broadly, our study offers evidence-based insights to improve the peer review process, guiding authors on effective rebuttal strategies and helping the community design fairer and more efficient review processes. Our code and score changes data are available at https://github.com/papercopilot/iclr-insights.
Authors: Rashid Iqbal, Saddam Hussain Khan
Abstract: This work proposes a hybrid deep learning approach, namely Residual and Spatial Learning based Channel Augmented Integrated CNN-Transformer architecture, that leverages the strengths of CNN and Transformer towards enhanced MPox detection. The proposed RS-CA-HSICT framework is composed of an HSICT block, a residual CNN module, a spatial CNN block, and a CA, which enhances the diverse feature space, detailed lesion information, and long-range dependencies. The new HSICT module first integrates an abstract representation of the stem CNN and customized ICT blocks for efficient multihead attention and structured CNN layers with homogeneous (H) and structural (S) operations. The customized ICT blocks learn global contextual interactions and local texture extraction. Additionally, H and S layers learn spatial homogeneity and fine structural details by reducing noise and modeling complex morphological variations. Moreover, inverse residual learning enhances vanishing gradient, and stage-wise resolution reduction ensures scale invariance. Furthermore, the RS-CA-HSICT framework augments the learned HSICT channels with the TL-driven Residual and Spatial CNN maps for enhanced multiscale feature space capturing global and localized structural cues, subtle texture, and contrast variations. These channels, preceding augmentation, are refined through the Channel-Fusion-and-Attention block, which preserves discriminative channels while suppressing redundant ones, thereby enabling efficient computation. Finally, the spatial attention mechanism refines pixel selection to detect subtle patterns and intra-class contrast variations in Mpox. Experimental results on both the Kaggle benchmark and a diverse MPox dataset reported classification accuracy as high as 98.30% and an F1-score of 98.13%, which outperforms the existing CNNs and ViTs.
Authors: Maria Pilligua, David Serrano-Lozano, Pai Peng, Ramon Baldrich, Michael S. Brown, Javier Vazquez-Corral
Abstract: Imaging in low-light environments is challenging due to reduced scene radiance, which leads to elevated sensor noise and reduced color saturation. Most learning-based low-light enhancement methods rely on paired training data captured under a single low-light condition and a well-lit reference. The lack of radiance diversity limits our understanding of how enhancement techniques perform across varying illumination intensities. We introduce the Multi-Illumination Low-Light (MILL) dataset, containing images captured at diverse light intensities under controlled conditions with fixed camera settings and precise illuminance measurements. MILL enables comprehensive evaluation of enhancement algorithms across variable lighting conditions. We benchmark several state-of-the-art methods and reveal significant performance variations across intensity levels. Leveraging the unique multi-illumination structure of our dataset, we propose improvements that enhance robustness across diverse illumination scenarios. Our modifications achieve up to 10 dB PSNR improvement for DSLR and 2 dB for the smartphone on Full HD images.
Authors: Gabriel Lauzier, Alexandre Girard, Fran\c{c}ois Ferland
Abstract: Diffusion Policy has shown great performance in robotic manipulation tasks under stochastic perturbations, due to its ability to model multimodal action distributions. Nonetheless, its reliance on a computationally expensive reverse-time diffusion (denoising) process, for action inference, makes it challenging to use for real-time applications where quick decision-making is mandatory. This work studies the possibility of conducting the denoising process only partially before executing an action, allowing the plant to evolve according to its dynamics in parallel to the reverse-time diffusion dynamics ongoing on the computer. In a classical diffusion policy setting, the plant dynamics are usually slow and the two dynamical processes are uncoupled. Here, we investigate theoretical bounds on the stability of closed-loop systems using diffusion policies when the plant dynamics and the denoising dynamics are coupled. The contribution of this work gives a framework for faster imitation learning and a metric that yields if a controller will be stable based on the variance of the demonstrations.
Authors: Artem Chervyakov, Ulyana Isaeva, Anton Emelyanov, Artem Safin, Maria Tikhonova, Alexander Kharitonov, Yulia Lyakh, Petr Surovtsev, Denis Shevelev Vildan Saburov, Vasily Konovalov, Elisei Rykov, Ivan Sviridov, Amina Miftakhova, Ilseyar Alimova, Alexander Panchenko, Alexander Kapitanov, Alena Fenogenova
Abstract: Multimodal large language models (MLLMs) are currently at the center of research attention, showing rapid progress in scale and capabilities, yet their intelligence, limitations, and risks remain insufficiently understood. To address these issues, particularly in the context of the Russian language, where no multimodal benchmarks currently exist, we introduce Mera Multi, an open multimodal evaluation framework for Russian-spoken architectures. The benchmark is instruction-based and encompasses default text, image, audio, and video modalities, comprising 18 newly constructed evaluation tasks for both general-purpose models and modality-specific architectures (image-to-text, video-to-text, and audio-to-text). Our contributions include: (i) a universal taxonomy of multimodal abilities; (ii) 18 datasets created entirely from scratch with attention to Russian cultural and linguistic specificity, unified prompts, and metrics; (iii) baseline results for both closed-source and open-source models; (iv) a methodology for preventing benchmark leakage, including watermarking and licenses for private sets. While our current focus is on Russian, the proposed benchmark provides a replicable methodology for constructing multimodal benchmarks in typologically diverse languages, particularly within the Slavic language family.
Authors: Selim Furkan Tekin, Rajesh Bordawekar
Abstract: Storing and processing of embedding vectors by specialized Vector databases (VDBs) has become the linchpin in building modern AI pipelines. Most current VDBs employ variants of a graph-based ap- proximate nearest-neighbor (ANN) index algorithm, HNSW, to an- swer semantic queries over stored vectors. Inspite of its wide-spread use, the HNSW algorithm suffers from several issues: in-memory design and implementation, random memory accesses leading to degradation in cache behavior, limited acceleration scope due to fine-grained pairwise computations, and support of only semantic similarity queries. In this paper, we present a novel disk-based ANN index, B+ANN, to address these issues: it first partitions input data into blocks containing semantically similar items, then builds an B+ tree variant to store blocks both in-memory and on disks, and finally, enables hybrid edge- and block-based in-memory traversals. As demonstrated by our experimantal evaluation, the proposed B+ANN disk-based index improves both quality (Recall value), and execution performance (Queries per second/QPS) over HNSW, by improving spatial and temporal locality for semantic operations, reducing cache misses (19.23% relative gain), and decreasing the memory consumption and disk-based build time by 24x over the DiskANN algorithm. Finally, it enables dissimilarity queries, which are not supported by similarity-oriented ANN indices.
Authors: Qihao Yang, Xuelin Wang, Jiale Chen, Xuelian Dong, Yuxin Hao, Tianyong Hao
Abstract: Language acquisition is vital to revealing the nature of human language intelligence and has recently emerged as a promising perspective for improving the interpretability of large language models (LLMs). However, it is ethically and practically infeasible to conduct experiments that require controlling human learners' language inputs. This poses challenges for the verifiability and scalability of language acquisition modeling, particularly in Chinese second language acquisition (SLA). While LLMs provide a controllable and reproducible alternative, a systematic benchmark to support phase-wise modeling and assessment is still lacking. In this paper, we present HSKBenchmark, the first benchmark for staged modeling and writing assessment of LLMs in Chinese SLA. It covers HSK levels 3 to 6 and includes authentic textbooks with 6.76 million tokens, 16K synthetic instruction samples, 30 test topics, and a linguistically grounded evaluation system. To simulate human learning trajectories, we introduce a curriculum-tuning framework that trains models from beginner to advanced levels. An evaluation system is created to examine level-based grammar coverage, writing errors, lexical and syntactic complexity, and holistic scoring. We also build HSKAgent, fine-tuned on 10K learner compositions. Extensive experimental results demonstrate that HSKBenchmark not only models Chinese SLA effectively, but also serves as a reliable benchmark for dynamic writing assessment in LLMs. Our fine-tuned LLMs have writing performance on par with advanced human learners and exhibit human-like acquisition characteristics. The HSKBenchmark, HSKAgent, and checkpoints serve as foundational tools and resources, with the potential to pave the way for future research on language acquisition modeling and LLMs interpretability. Code and data are publicly available at: https://github.com/CharlesYang030/HSKB.
Authors: Sifan Zhou, Yichao Cao, Jiahao Nie, Yuqian Fu, Ziyu Zhao, Xiaobo Lu, Shuo Wang
Abstract: 3D single object tracking (SOT) in LiDAR point clouds is a critical task in computer vision and autonomous driving. Despite great success having been achieved, the inherent sparsity of point clouds introduces a dual-redundancy challenge that limits existing trackers: (1) vast spatial redundancy from background noise impairs accuracy, and (2) informational redundancy within the foreground hinders efficiency. To tackle these issues, we propose CompTrack, a novel end-to-end framework that systematically eliminates both forms of redundancy in point clouds. First, CompTrack incorporates a Spatial Foreground Predictor (SFP) module to filter out irrelevant background noise based on information entropy, addressing spatial redundancy. Subsequently, its core is an Information Bottleneck-guided Dynamic Token Compression (IB-DTC) module that eliminates the informational redundancy within the foreground. Theoretically grounded in low-rank approximation, this module leverages an online SVD analysis to adaptively compress the redundant foreground into a compact and highly informative set of proxy tokens. Extensive experiments on KITTI, nuScenes and Waymo datasets demonstrate that CompTrack achieves top-performing tracking performance with superior efficiency, running at a real-time 90 FPS on a single RTX 3090 GPU.
Authors: Sai Puppala, Ismail Hossain, Jahangir Alam, Sajedul Talukder
Abstract: The integration of advanced robotics in nuclear power plants (NPPs) presents a transformative opportunity to enhance safety, efficiency, and environmental monitoring in high-stakes environments. Our paper introduces the Optimus-Q robot, a sophisticated system designed to autonomously monitor air quality and detect contamination while leveraging adaptive learning techniques and secure quantum communication. Equipped with advanced infrared sensors, the Optimus-Q robot continuously streams real-time environmental data to predict hazardous gas emissions, including carbon dioxide (CO$_2$), carbon monoxide (CO), and methane (CH$_4$). Utilizing a federated learning approach, the robot collaborates with other systems across various NPPs to improve its predictive capabilities without compromising data privacy. Additionally, the implementation of Quantum Key Distribution (QKD) ensures secure data transmission, safeguarding sensitive operational information. Our methodology combines systematic navigation patterns with machine learning algorithms to facilitate efficient coverage of designated areas, thereby optimizing contamination monitoring processes. Through simulations and real-world experiments, we demonstrate the effectiveness of the Optimus-Q robot in enhancing operational safety and responsiveness in nuclear facilities. This research underscores the potential of integrating robotics, machine learning, and quantum technologies to revolutionize monitoring systems in hazardous environments.
Authors: Dante Francisco Wasmuht, Otto Brookes, Maximillian Schall, Pablo Palencia, Chris Beirne, Tilo Burghardt, Majid Mirmehdi, Hjalmar K\"uhl, Mimi Arandjelovic, Sam Pottie, Peter Bermant, Brandon Asheim, Yi Jin Toh, Adam Elzinga, Jason Holmberg, Andrew Whitworth, Eleanor Flatt, Laura Gustafson, Chaitanya Ryali, Yuan-Ting Hu, Baishan Guo, Andrew Westbury, Kate Saenko, Didac Suris
Abstract: Automated video analysis is critical for wildlife conservation. A foundational task in this domain is multi-animal tracking (MAT), which underpins applications such as individual re-identification and behavior recognition. However, existing datasets are limited in scale, constrained to a few species, or lack sufficient temporal and geographical diversity - leaving no suitable benchmark for training general-purpose MAT models applicable across wild animal populations. To address this, we introduce SA-FARI, the largest open-source MAT dataset for wild animals. It comprises 11,609 camera trap videos collected over approximately 10 years (2014-2024) from 741 locations across 4 continents, spanning 99 species categories. Each video is exhaustively annotated culminating in ~46 hours of densely annotated footage containing 16,224 masklet identities and 942,702 individual bounding boxes, segmentation masks, and species labels. Alongside the task-specific annotations, we publish anonymized camera trap locations for each video. Finally, we present comprehensive benchmarks on SA-FARI using state-of-the-art vision-language models for detection and tracking, including SAM 3, evaluated with both species-specific and generic animal prompts. We also compare against vision-only methods developed specifically for wildlife analysis. SA-FARI is the first large-scale dataset to combine high species diversity, multi-region coverage, and high-quality spatio-temporal annotations, offering a new foundation for advancing generalizable multianimal tracking in the wild. The dataset is available at $\href{https://www.conservationxlabs.com/sa-fari}{\text{conservationxlabs.com/SA-FARI}}$.
Authors: Leopoldo Bertossi, Nina Pardal
Abstract: The notion of cause, as formalized by Halpern and Pearl, has been recently applied to relational databases, to characterize and compute causal explanations for query answers. In this work we consider the alternative notion of sufficient explanation. We investigate its connections with database repairs as used for dealing with inconsistent databases, and with causality-based necessary explanations. We also obtain some computational results.
Authors: Kim N. Nolle, Ivana Dusparic, Rhodri Cusack, Vinny Cahill
Abstract: Continual learning (CL) is a branch of machine learning that aims to enable agents to adapt and generalise previously learned abilities so that these can be reapplied to new tasks or environments. This is particularly useful in multi-task settings or in non-stationary environments, where the dynamics can change over time. This is particularly relevant in cyber-physical systems such as autonomous driving. However, despite recent advances in CL, successfully applying it to reinforcement learning (RL) is still an open problem. This paper highlights open challenges in continual RL (CRL) based on experiments in an autonomous driving environment. In this environment, the agent must learn to successfully park in four different scenarios corresponding to parking spaces oriented at varying angles. The agent is successively trained in these four scenarios one after another, representing a CL environment, using Proximal Policy Optimisation (PPO). These experiments exposed a number of open challenges in CRL: finding suitable abstractions of the environment, oversensitivity to hyperparameters, catastrophic forgetting, and efficient use of neural network capacity. Based on these identified challenges, we present open research questions that are important to be addressed for creating robust CRL systems. In addition, the identified challenges call into question the suitability of neural networks for CL. We also identify the need for interdisciplinary research, in particular between computer science and neuroscience.
Authors: Naomi Simumba, Nils Lehmann, Paolo Fraccaro, Hamed Alemohammad, Geeth De Mel, Salman Khan, Manil Maskey, Nicolas Longepe, Xiao Xiang Zhu, Hannah Kerner, Juan Bernabe-Moreno, Alexander Lacoste
Abstract: Geospatial Foundation Models (GeoFMs) are transforming Earth Observation (EO), but evaluation lacks standardized protocols. GEO-Bench-2 addresses this with a comprehensive framework spanning classification, segmentation, regression, object detection, and instance segmentation across 19 permissively-licensed datasets. We introduce ''capability'' groups to rank models on datasets that share common characteristics (e.g., resolution, bands, temporality). This enables users to identify which models excel in each capability and determine which areas need improvement in future work. To support both fair comparison and methodological innovation, we define a prescriptive yet flexible evaluation protocol. This not only ensures consistency in benchmarking but also facilitates research into model adaptation strategies, a key and open challenge in advancing GeoFMs for downstream tasks. Our experiments show that no single model dominates across all tasks, confirming the specificity of the choices made during architecture design and pretraining. While models pretrained on natural images (ConvNext ImageNet, DINO V3) excel on high-resolution tasks, EO-specific models (TerraMind, Prithvi, and Clay) outperform them on multispectral applications such as agriculture and disaster response. These findings demonstrate that optimal model choice depends on task requirements, data modalities, and constraints. This shows that the goal of a single GeoFM model that performs well across all tasks remains open for future research. GEO-Bench-2 enables informed, reproducible GeoFM evaluation tailored to specific use cases. Code, data, and leaderboard for GEO-Bench-2 are publicly released under a permissive license.
Authors: Yicheng He, Chengsong Huang, Zongxia Li, Jiaxin Huang, Yonghui Yang
Abstract: Reinforcement learning (RL) provides a principled framework for improving Vision-Language Models (VLMs) on complex reasoning tasks. However, existing RL approaches often rely on human-annotated labels or task-specific heuristics to define verifiable rewards, both of which are costly and difficult to scale. We introduce VisPlay, a self-evolving RL framework that enables VLMs to autonomously improve their reasoning abilities using large amounts of unlabeled image data. Starting from a single base VLM, VisPlay assigns the model into two interacting roles: an Image-Conditioned Questioner that formulates challenging yet answerable visual questions, and a Multimodal Reasoner that generates silver responses. These roles are jointly trained with Group Relative Policy Optimization (GRPO), which incorporates diversity and difficulty rewards to balance the complexity of generated questions with the quality of the silver answers. VisPlay scales efficiently across two model families. When trained on Qwen2.5-VL and MiMo-VL, VisPlay achieves consistent improvements in visual reasoning, compositional generalization, and hallucination reduction across eight benchmarks, including MM-Vet and MMMU, demonstrating a scalable path toward self-evolving multimodal intelligence. The project page is available at https://bruno686.github.io/VisPlay/
Authors: Cheng Yin, Yankai Lin, Wang Xu, Sikyuen Tam, Xiangrui Zeng, Zhiyuan Liu, Zhouping Yin
Abstract: Enabling Vision-Language-Action (VLA) models to "think before acting" via Chain-of-Thought (CoT) is a promising path to overcoming the data-hungry nature of end-to-end robot policies. However, progress is stalled by a fundamental conflict: existing models use a single autoregressive decoder for both sequential CoT reasoning and high-dimensional, parallelizable robot actions. This architectural mismatch degrades motor control and fails to forge a strong causal link between thought and action. We introduce DeepThinkVLA, which resolves this conflict through a tightly integrated architecture and training strategy. Architecturally, our hybrid-attention decoder generates sequential CoT with causal attention and then switches to bidirectional attention for fast, parallel decoding of action vectors. This design is complemented by a two-stage training pipeline: we first use Supervised Fine-Tuning (SFT) to teach the model foundational reasoning, then apply Reinforcement Learning (RL) with task-success rewards to causally align the full reasoning-action sequence with desired outcomes. This synergy leads to state-of-the-art performance, achieving a 97.0% success rate on the LIBERO benchmark. Our ablations confirm the design's effectiveness: the hybrid architecture alone outperforms standard decoders by 15.5%, and the final RL stage provides a crucial 2% boost to secure top performance.
Authors: Sejuti Rahman, Swakshar Deb, MD. Sameer Iqbal Chowdhury, MD. Jubair Ahmed Sourov, Mohammad Shamsuddin
Abstract: Eye tracking data quantifies the attentional bias towards negative stimuli that is frequently observed in depressed groups. Audio and video data capture the affective flattening and psychomotor retardation characteristic of depression. Statistical validation confirmed their significant discriminative power in distinguishing depressed from non depressed groups. We address a critical limitation of existing graph-based models that focus on low-frequency information and propose a Multi-Frequency Graph Convolutional Network (MF-GCN). This framework consists of a novel Multi-Frequency Filter Bank Module (MFFBM), which can leverage both low and high frequency signals. Extensive evaluation against traditional machine learning algorithms and deep learning frameworks demonstrates that MF-GCN consistently outperforms baselines. In binary (depressed and non depressed) classification, the model achieved a sensitivity of 0.96 and F2 score of 0.94. For the 3 class (no depression, mild to moderate depression and severe depression) classification task, the proposed method achieved a sensitivity of 0.79 and specificity of 0.87 and siginificantly suprassed other models. To validate generalizability, the model was also evaluated on the Chinese Multimodal Depression Corpus (CMDC) dataset and achieved a sensitivity of 0.95 and F2 score of 0.96. These results confirm that our trimodal, multi frequency framework effectively captures cross modal interaction for accurate depression detection.
Authors: Michael McCabe, Payel Mukhopadhyay, Tanya Marwah, Bruno Regaldo-Saint Blancard, Francois Rozet, Cristiana Diaconu, Lucas Meyer, Kaze W. K. Wong, Hadi Sotoudeh, Alberto Bietti, Irina Espejo, Rio Fear, Siavash Golkar, Tom Hehir, Keiya Hirashima, Geraud Krawezik, Francois Lanusse, Rudy Morel, Ruben Ohana, Liam Parker, Mariel Pettee, Jeff Shen, Kyunghyun Cho, Miles Cranmer, Shirley Ho
Abstract: Foundation models have transformed machine learning for language and vision, but achieving comparable impact in physical simulation remains a challenge. Data heterogeneity and unstable long-term dynamics inhibit learning from sufficiently diverse dynamics, while varying resolutions and dimensionalities challenge efficient training on modern hardware. Through empirical and theoretical analysis, we incorporate new approaches to mitigate these obstacles, including a harmonic-analysis-based stabilization method, load-balanced distributed 2D and 3D training strategies, and compute-adaptive tokenization. Using these tools, we develop Walrus, a transformer-based foundation model developed primarily for fluid-like continuum dynamics. Walrus is pretrained on nineteen diverse scenarios spanning astrophysics, geoscience, rheology, plasma physics, acoustics, and classical fluids. Experiments show that Walrus outperforms prior foundation models on both short and long term prediction horizons on downstream tasks and across the breadth of pretraining data, while ablation studies confirm the value of our contributions to forecast stability, training throughput, and transfer performance over conventional approaches. Code and weights are released for community use.
Authors: Jingkai Ying, Zhijin Qin, Yulong Feng, Liejun Wang, Xiaoming Tao
Abstract: In recent years, the Transformer architecture has achieved outstanding performance across a wide range of tasks and modalities. Token is the unified input and output representation in Transformer-based models, which has become a fundamental information unit. In this work, we consider the problem of token communication, studying how to transmit tokens efficiently and reliably. Point cloud, a prevailing three-dimensional format which exhibits a more complex spatial structure compared to image or video, is chosen to be the information source. We utilize the set abstraction method to obtain point tokens. Subsequently, to get a more informative and transmission-friendly representation based on tokens, we propose a joint semantic-channel and modulation (JSCCM) scheme for the token encoder, mapping point tokens to standard digital constellation points (modulated tokens). Specifically, the JSCCM consists of two parallel Point Transformer-based encoders and a differential modulator which combines the Gumel-softmax and soft quantization methods. Besides, the rate allocator and channel adapter are developed, facilitating adaptive generation of high-quality modulated tokens conditioned on both semantic information and channel conditions. Extensive simulations demonstrate that the proposed method outperforms both joint semantic-channel coding and traditional separate coding, achieving over 1dB gain in reconstruction and more than 6x compression ratio in modulated symbols.
Authors: Beichen Zhang, Yuhang Zang, Xiaoyi Dong, Yuhang Cao, Haodong Duan, Dahua Lin, Jiaqi Wang
Abstract: Abstract reasoning from minimal examples remains a core unsolved problem for frontier foundation models such as GPT-5 and Grok 4. These models still fail to infer structured transformation rules from a handful of examples, which is a key hallmark of human intelligence. The Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI) provides a rigorous testbed for this capability, demanding conceptual rule induction and transfer to novel tasks. Most existing methods treat ARC-AGI as a purely textual reasoning task, overlooking the fact that humans rely heavily on visual abstraction when solving such puzzles. However, our pilot experiments reveal a paradox: naively rendering ARC-AGI grids as images degrades performance due to imprecise rule execution. This leads to our central hypothesis that vision and language possess complementary strengths across distinct reasoning stages: vision supports global pattern abstraction and verification, whereas language specializes in symbolic rule formulation and precise execution. Building on this insight, we introduce two synergistic strategies: (1) Vision-Language Synergy Reasoning (VLSR), which decomposes ARC-AGI into modality-aligned subtasks; and (2) Modality-Switch Self-Correction (MSSC), which leverages vision to verify text-based reasoning for intrinsic error correction. Extensive experiments demonstrate that our approach yields up to a 4.33% improvement over text-only baselines across diverse flagship models and multiple ARC-AGI tasks. Our findings suggest that unifying visual abstraction with linguistic reasoning is a crucial step toward achieving generalizable, human-like intelligence in future foundation models. Source code will be released soon.
Authors: Xiongyi Cai, Ri-Zhao Qiu, Geng Chen, Lai Wei, Isabella Liu, Tianshu Huang, Xuxin Cheng, Xiaolong Wang
Abstract: Egocentric videos are a valuable and scalable data source to learn manipulation policies. However, due to significant data heterogeneity, most existing approaches utilize human data for simple pre-training, which does not unlock its full potential. This paper first provides a scalable recipe for collecting and using egocentric data by categorizing human data into two categories: in-the-wild and on-task alongside with systematic analysis on how to use the data. We first curate a dataset, PHSD, which contains over 1,000 hours of diverse in-the-wild egocentric data and over 20 hours of on-task data directly aligned to the target manipulation tasks. This enables learning a large egocentric language-conditioned flow matching policy, Human0. With domain adaptation techniques, Human0 minimizes the gap between humans and humanoids. Empirically, we show Human0 achieves several novel properties from scaling human data, including language following of instructions from only human data, few-shot learning, and improved robustness using on-task data. Project website: https://xiongyicai.github.io/In-N-On/
Authors: Tianhui Cai, Yifan Liu, Zewei Zhou, Haoxuan Ma, Seth Z. Zhao, Zhiwen Wu, Xu Han, Zhiyu Huang, Jiaqi Ma
Abstract: Understanding and adhering to traffic regulations is essential for autonomous vehicles to ensure safety and trustworthiness. However, traffic regulations are complex, context-dependent, and differ between regions, posing a major challenge to conventional rule-based decision-making approaches. We present an interpretable, regulation-aware decision-making framework, DriveReg, which enables autonomous vehicles to understand and adhere to region-specific traffic laws and safety guidelines. The framework integrates a Retrieval-Augmented Generation (RAG)-based Traffic Regulation Retrieval Agent, which retrieves relevant rules from regulatory documents based on the current situation, and a Large Language Model (LLM)-powered Reasoning Agent that evaluates actions for legal compliance and safety. Our design emphasizes interpretability to enhance transparency and trustworthiness. To support systematic evaluation, we introduce the DriveReg Scenarios Dataset, a comprehensive dataset of driving scenarios across Boston, Singapore, and Los Angeles, with both hypothesized text-based cases and real-world driving data, constructed and annotated to evaluate models' capacity for regulation understanding and reasoning. We validate our framework on the DriveReg Scenarios Dataset and real-world deployment, demonstrating strong performance and robustness across diverse environments.
Authors: Yu Li, Yi Huang, Guilin Qi, Junlan Feng, Nan Hu, Songlin Zhai, Haohan Xue, Yongrui Chen, Ruoyan Shen, Tongtong Wu
Abstract: Knowledge graphs are widely used in industrial applications, making error detection crucial for ensuring the reliability of downstream applications. Existing error detection methods often fail to effectively utilize fine-grained subgraph information and rely solely on fixed graph structures, while also lacking transparency in their decision-making processes, which results in suboptimal detection performance. In this paper, we propose a novel Multi-Agent framework for Knowledge Graph Error Detection (MAKGED) that utilizes multiple large language models (LLMs) in a collaborative setting. By concatenating fine-grained, bidirectional subgraph embeddings with LLM-based query embeddings during training, our framework integrates these representations to produce four specialized agents. These agents utilize subgraph information from different dimensions to engage in multi-round discussions, thereby improving error detection accuracy and ensuring a transparent decision-making process. Extensive experiments on FB15K and WN18RR demonstrate that MAKGED outperforms state-of-the-art methods, enhancing the accuracy and robustness of KG evaluation. For specific industrial scenarios, our framework can facilitate the training of specialized agents using domain-specific knowledge graphs for error detection, which highlights the potential industrial application value of our framework. Our code and datasets are available at https://github.com/kse-ElEvEn/MAKGED.
Authors: Aran Nayebi
Abstract: We formalize AI alignment as a multi-objective optimization problem called $\langle M,N,\varepsilon,\delta\rangle$-agreement, in which a set of $N$ agents (including humans) must reach approximate ($\varepsilon$) agreement across $M$ candidate objectives, with probability at least $1-\delta$. Analyzing communication complexity, we prove an information-theoretic lower bound showing that once either $M$ or $N$ is large enough, no amount of computational power or rationality can avoid intrinsic alignment overheads. This establishes rigorous limits to alignment *itself*, not merely to particular methods, clarifying a "No-Free-Lunch" principle: encoding "all human values" is inherently intractable and must be managed through consensus-driven reduction or prioritization of objectives. Complementing this impossibility result, we construct explicit algorithms as achievability certificates for alignment under both unbounded and bounded rationality with noisy communication. Even in these best-case regimes, our bounded-agent and sampling analysis shows that with large task spaces ($D$) and finite samples, *reward hacking is globally inevitable*: rare high-loss states are systematically under-covered, implying scalable oversight must target safety-critical slices rather than uniform coverage. Together, these results identify fundamental complexity barriers -- tasks ($M$), agents ($N$), and state-space size ($D$) -- and offer principles for more scalable human-AI collaboration.
Authors: Weixiang Zhao, Xingyu Sui, Jiahe Guo, Yulin Hu, Yang Deng, Yanyan Zhao, Xuda Zhi, Yongbo Huang, Hao He, Wanxiang Che, Ting Liu, Bing Qin
Abstract: Recent advancements in Large Reasoning Models (LRMs), such as OpenAI's o1/o3 and DeepSeek-R1, have demonstrated remarkable performance in specialized reasoning tasks through human-like deliberative thinking and long chain-of-thought reasoning. However, our systematic evaluation across various model families (DeepSeek, Qwen, and LLaMA) and scales (7B to 32B) reveals that acquiring these deliberative reasoning capabilities significantly reduces the foundational capabilities of LRMs, including notable declines in helpfulness and harmlessness, alongside substantially increased inference costs. Importantly, we demonstrate that adaptive reasoning -- employing modes like Zero-Thinking, Less-Thinking, and Summary-Thinking -- can effectively alleviate these drawbacks. Our empirical insights underline the critical need for developing more versatile LRMs capable of dynamically allocating inference-time compute according to specific task characteristics.
Authors: Linqiang Guo (Peter), Wei Liu (Peter), Yi Wen Heng (Peter), Tse-Hsun (Peter), Chen, Yang Wang
Abstract: Mobile Graphical User Interface (GUI) agents aim to autonomously complete tasks within or across apps based on user instructions. While recent Multimodal Large Language Models (MLLMs) enable these agents to interpret UI screens and perform actions, existing agents remain fundamentally reactive. They reason over the current UI screen but lack a structured representation of the app navigation flow, limiting GUI agents' ability to understand execution context, detect unexpected execution results, and recover from errors. We introduce Agent-SAMA, a state-aware multi-agent framework that models app execution as a Finite State Machine (FSM), treating UI screens as states and user actions as transitions. Agent-SAMA implements four specialized agents that collaboratively construct and use FSMs in real time to guide task planning, execution verification, and recovery. We evaluate Agent-SAMA on two types of benchmarks: cross-app (Mobile-Eval-E, SPA-Bench) and mostly single-app (AndroidWorld). On Mobile-Eval-E, Agent-SAMA achieves an 84.0% success rate and a 71.9% recovery rate. On SPA-Bench, it reaches an 80.0% success rate with a 66.7% recovery rate. Compared to prior methods, Agent-SAMA improves task success by up to 12% and recovery success by 13.8%. On AndroidWorld, Agent-SAMA achieves a 63.7% success rate, outperforming the baselines. Our results demonstrate that structured state modeling enhances robustness and can serve as a lightweight, model-agnostic memory layer for future GUI agents.
Authors: Joaqu\'in Jord\'an, Xavier Yin, Melissa Fabros, Gireeja Ranade, Narges Norouzi
Abstract: Automated Essay Scoring (AES) and Automatic Essay Feedback (AEF) systems aim to reduce the workload of human raters in educational assessment. However, most existing systems prioritize numerical scoring accuracy over feedback quality and are primarily evaluated on pre-secondary school level writing. This paper presents Multi-Agent Argumentation and Grammar Integrated Critiquer (MAGIC), a framework using five specialized agents to evaluate prompt adherence, persuasiveness, organization, vocabulary, and grammar for both holistic scoring and detailed feedback generation. To support evaluation at the college level, we collated a dataset of Graduate Record Examination (GRE) practice essays with expert-evaluated scores and feedback. MAGIC achieves substantial to near-perfect scoring agreement with humans on the GRE data, outperforming baseline LLM models while providing enhanced interpretability through its multi-agent approach. We also compare MAGIC's feedback generation capabilities against ground truth human feedback and baseline models, finding that MAGIC achieves strong feedback quality and naturalness.
Authors: Aran Nayebi
Abstract: We introduce the first complete formal solution to corrigibility in the off-switch game, with provable guarantees in multi-step, partially observed environments. Our framework consists of five *structurally separate* utility heads -- deference, switch-access preservation, truthfulness, low-impact behavior via a belief-based extension of Attainable Utility Preservation, and bounded task reward -- combined lexicographically by strict weight gaps. Theorem 1 proves exact single-round corrigibility in the partially observable off-switch game; Theorem 3 extends the guarantee to multi-step, self-spawning agents, showing that even if each head is *learned* to mean-squared error $\varepsilon$ and the planner is $\varepsilon$-sub-optimal, the probability of violating *any* safety property is bounded while still ensuring net human benefit. In contrast to Constitutional AI or RLHF/RLAIF, which merge all norms into one learned scalar, our separation makes obedience and impact-limits provably dominate even when incentives conflict. For settings where adversaries can modify the agent, we prove that deciding whether an arbitrary post-hack agent will ever violate corrigibility is undecidable by reduction to the halting problem, then carve out a finite-horizon "decidable island" where safety can be certified in randomized polynomial time and verified with privacy-preserving, constant-round zero-knowledge proofs.
Authors: Alessandro Abate, Thom Badings, Giuseppe De Giacomo, Francesco Fabiano
Abstract: We study the common generalization of Markov decision processes (MDPs) with sets of transition probabilities, known as robust MDPs (RMDPs). A standard goal in RMDPs is to compute a policy that maximizes the expected return under an adversarial choice of the transition probabilities. If the uncertainty in the probabilities is independent between the states, known as s-rectangularity, such optimal robust policies can be computed efficiently using robust value iteration. However, there might still be multiple optimal robust policies, which, while equivalent with respect to the worst-case, reflect different expected returns under non-adversarial choices of the transition probabilities. Hence, we propose a refined policy selection criterion for RMDPs, drawing inspiration from the notions of dominance and best-effort in game theory. Instead of seeking a policy that only maximizes the worst-case expected return, we additionally require the policy to achieve a maximal expected return under different (i.e., not fully adversarial) transition probabilities. We call such a policy an optimal robust best-effort (ORBE) policy. We prove that ORBE policies always exist, characterize their structure, and present an algorithm to compute them with a manageable overhead compared to standard robust value iteration. ORBE policies offer a principled tie-breaker among optimal robust policies. Numerical experiments show the feasibility of our approach.
Authors: Guoying Zhu, Meng Li, Haipeng Dai, Xuechen Liu, Weijun Wang, Keran Li, Jun xiao, Ligeng Chen, Wei Wang
Abstract: The Mixture of Experts (MoE) architecture has emerged as a key technique for scaling Large Language Models by activating only a subset of experts per query. Deploying MoE on consumer-grade edge hardware, however, is constrained by limited device memory, making dynamic expert offloading essential. Unlike prior work that treats offloading purely as a scheduling problem, we leverage expert importance to guide decisions, substituting low-importance activated experts with functionally similar ones already cached in GPU memory, thereby preserving accuracy. As a result, this design reduces memory usage and data transfer, while largely eliminating PCIe overhead. In addition, we introduce a scheduling policy that maximizes the reuse ratio of GPU-cached experts, further boosting efficiency. Extensive evaluations show that our approach delivers 48% lower decoding latency with over 60% expert cache hit rate, while maintaining nearly lossless accuracy.
Authors: Siqi Ma, Jiajie Huang, Fan Zhang, Jinlin Wu, Yue Shen, Guohui Fan, Zhu Zhang, Zelin Zang
Abstract: Answering complex medical questions requires not only domain expertise and patient-specific information, but also structured and multi-perspective reasoning. Existing multi-agent approaches often rely on fixed roles or shallow interaction prompts, limiting their ability to detect and resolve fine-grained logical inconsistencies. To address this, we propose \textsc{MedLA}, a logic-driven multi-agent framework built on large language models. Each agent organizes its reasoning process into an explicit logical tree based on syllogistic triads (major premise, minor premise, and conclusion), enabling transparent inference and premise-level alignment. Agents engage in a multi-round, graph-guided discussion to compare and iteratively refine their logic trees, achieving consensus through error correction and contradiction resolution. We demonstrate that \textsc{MedLA} consistently outperforms both static role-based systems and single-agent baselines on challenging benchmarks such as MedDDx and standard medical QA tasks. Furthermore, \textsc{MedLA} scales effectively across both open-source and commercial LLM backbones, achieving state-of-the-art performance and offering a generalizable paradigm for trustworthy medical reasoning.
Authors: Fei Yang
Abstract: Human prowess in intuitive physics remains unmatched by machines. To bridge this gap, we argue for a fundamental shift towards brain-inspired computational principles. This paper introduces the Spatiotemporal Relational Neural Network (SRNN), a model that establishes a unified neural representation for object attributes, relations, and timeline, with computations governed by a Hebbian ``Fire Together, Wire Together'' mechanism across dedicated \textit{What} and \textit{How} pathways. This unified representation is directly used to generate structured linguistic descriptions of the visual scene, bridging perception and language within a shared neural substrate. On the CLEVRER benchmark, SRNN achieves competitive performance, thereby confirming its capability to represent essential spatiotemporal relations from the visual stream. Cognitive ablation analysis further reveals a benchmark bias, outlining a path for a more holistic evaluation. Finally, the white-box nature of SRNN enables precise pinpointing of error root causes. Our work provides a proof-of-concept that confirms the viability of translating key principles of biological intelligence into engineered systems for intuitive physics understanding in constrained environments.
Authors: He Panjing, Cheng Mingyue, Li Li, Zhang XiaoHan
Abstract: Generating high-quality time series data has emerged as a critical research topic due to its broad utility in supporting downstream time series mining tasks. A major challenge lies in modeling the intrinsic stochasticity of temporal dynamics, as real-world sequences often exhibit random fluctuations and localized variations. While diffusion models have achieved remarkable success, their generation process is computationally inefficient, often requiring hundreds to thousands of expensive function evaluations per sample. Flow matching has emerged as a more efficient paradigm, yet its conventional ordinary differential equation (ODE)-based formulation fails to explicitly capture stochasticity, thereby limiting the fidelity of generated sequences. By contrast, stochastic differential equation (SDE) are naturally suited for modeling randomness and uncertainty. Motivated by these insights, we propose TimeFlow, a novel SDE-based flow matching framework that integrates a encoder-only architecture. Specifically, we design a component-wise decomposed velocity field to capture the multi-faceted structure of time series and augment the vanilla flow-matching optimization with an additional stochastic term to enhance representational expressiveness. TimeFlow is flexible and general, supporting both unconditional and conditional generation tasks within a unified framework. Extensive experiments across diverse datasets demonstrate that our model consistently outperforms strong baselines in generation quality, diversity, and efficiency.
Authors: Wenhan Yu, Xinbo Lin, Lanxin Ni, Jinhua Cheng, Lei Sha
Abstract: Large language models (LLMs) have demonstrated strong reasoning abilities across specialized domains, motivating research into their application to legal reasoning. However, existing legal benchmarks often conflate factual recall with genuine inference, fragment the reasoning process, and overlook the quality of reasoning. To address these limitations, we introduce MSLR, the first Chinese multi-step legal reasoning dataset grounded in real-world judicial decision making. MSLR adopts the IRAC framework (Issue, Rule, Application, Conclusion) to model structured expert reasoning from official legal documents. In addition, we design a scalable Human-LLM collaborative annotation pipeline that efficiently produces fine-grained step-level reasoning annotations and provides a reusable methodological framework for multi-step reasoning datasets. Evaluation of multiple LLMs on MSLR shows only moderate performance, highlighting the challenges of adapting to complex legal reasoning. Further experiments demonstrate that Self-Initiated Chain-of-Thought prompts generated by models autonomously improve reasoning coherence and quality, outperforming human-designed prompts. MSLR contributes to advancing LLM reasoning and Chain-of-Thought strategies and offers open resources for future research. The dataset and code are available at https://github.com/yuwenhan07/MSLR-Bench and https://law.sjtu.edu.cn/flszyjzx/index.html.
URLs: https://github.com/yuwenhan07/MSLR-Bench, https://law.sjtu.edu.cn/flszyjzx/index.html.
Authors: Zhiqi Chen, Yuzhou Liu, Jiarui Liu, Wanfu Gao
Abstract: Multi-view multi-label feature selection aims to identify informative features from heterogeneous views, where each sample is associated with multiple interdependent labels. This problem is particularly important in machine learning involving high-dimensional, multimodal data such as social media, bioinformatics or recommendation systems. Existing Multi-View Multi-Label Feature Selection (MVMLFS) methods mainly focus on analyzing statistical information of data, but seldom consider semantic information. In this paper, we aim to use these two types of information jointly and propose a method that combines Large Language Models (LLMs) semantic reasoning with Graph Neural Networks (GNNs) structural modeling for MVMLFS. Specifically, the method consists of three main components. (1) LLM is first used as an evaluation agent to assess the latent semantic relevance among feature, view, and label descriptions. (2) A semantic-aware heterogeneous graph with two levels is designed to represent relations among features, views and labels: one is a semantic graph representing semantic relations, and the other is a statistical graph. (3) A lightweight Graph Attention Network (GAT) is applied to learn node embedding in the heterogeneous graph as feature saliency scores for ranking and selection. Experimental results on multiple benchmark datasets demonstrate the superiority of our method over state-of-the-art baselines, and it is still effective when applied to small-scale datasets, showcasing its robustness, flexibility, and generalization ability.
Authors: Yaodong Yang, Yang Wang, Jinpeng Li, Pei Guo, Da Han, Guangyong Chen, Pheng-Ann Heng
Abstract: Protein evolution through amino acid sequence mutations is a cornerstone of life sciences. While current in-silicon directed evolution algorithms largely focus on designing heuristic search strategies, they overlook how to integrate the transformative protein language models, which encode rich evolutionary patterns, with reinforcement learning to learn to directly evolve proteins. To bridge this gap, we propose AlphaDE, a novel framework to optimize protein sequences by harnessing the innovative paradigms of large language models such as fine-tuning and test-time inference. First, AlphaDE fine-tunes pretrained protein language models using masked language modeling on homologous protein sequences to activate the evolutionary plausibility for the interested protein class. Second, AlphaDE introduces test-time inference based on Monte Carlo tree search, which effectively evolves proteins with evolutionary guidance from the fine-tuned protein language model. Extensive benchmark experiments show that AlphaDE remarkably outperforms previous state-of-the-art methods even with few-shot fine-tuning. A further case study demonstrates that AlphaDE supports condensing the protein sequence space of avGFP through computational evolution.
Authors: Yinghao Ruan, Wei Pang, Shuaihao Liu, Huili Yang, Leyi Han, Xinghui Dong
Abstract: The advent of smart manufacturing is addressing the limitations of traditional centralized scheduling and inflexible production line configurations in the rubber tyre industry, especially in terms of coping with dynamic production demands. Contemporary tyre manufacturing systems form complex networks of tightly coupled subsystems pronounced nonlinear interactions and emergent dynamics. This complexity renders the effective coordination of multiple subsystems, posing an essential yet formidable task. For high-dimensional, multi-objective optimization problems in this domain, we introduce a deep reinforcement learning algorithm: Multi-path Differentiated Clipping Proximal Policy Optimization (MPD-PPO). This algorithm employs a multi-branch policy architecture with differentiated gradient clipping constraints to ensure stable and efficient high-dimensional policy updates. Validated through experiments on width and thickness control in rubber tyre film production, MPD-PPO demonstrates substantial improvements in both tuning accuracy and operational efficiency. The framework successfully tackles key challenges, including high dimensionality, multi-objective trade-offs, and dynamic adaptation, thus delivering enhanced performance and production stability for real-time industrial deployment in tyre manufacturing.
Authors: Kaiyue Zhao, Dingqi Chen, Shaoyu Wang, Pan Hu
Abstract: DatalogMTL extends the classical Datalog language with metric temporal logic (MTL), enabling expressive reasoning over temporal data. While existing reasoning approaches, such as materialisation based and automata based methods, offer soundness and completeness, they lack support for handling efficient dynamic updates, a crucial requirement for real-world applications that involve frequent data updates. In this work, we propose DRedMTL, an incremental reasoning algorithm for DatalogMTL with bounded intervals. Our algorithm builds upon the classical DRed algorithm, which incrementally updates the materialisation of a Datalog program. Unlike a Datalog materialisation which is in essence a finite set of facts, a DatalogMTL materialisation has to be represented as a finite set of facts plus periodic intervals indicating how the full materialisation can be constructed through unfolding. To cope with this, our algorithm is equipped with specifically designed operators to efficiently handle such periodic representations of DatalogMTL materialisations. We have implemented this approach and tested it on several publicly available datasets. Experimental results show that DRedMTL often significantly outperforms rematerialisation, sometimes by orders of magnitude.
Authors: Baolong Bi, Shenghua Liu, Yiwei Wang, Siqian Tong, Lingrui Mei, Yuyao Ge, Yilong Xu, Jiafeng Guo, Xueqi Cheng
Abstract: Recent advances in reinforcement learning (RL) have significantly improved the complex reasoning capabilities of large language models (LLMs). Despite these successes, existing methods mainly focus on single-domain RL (e.g., mathematics) with verifiable rewards (RLVR), and their reliance on purely online RL frameworks restricts the exploration space, thereby limiting reasoning performance. In this paper, we address these limitations by leveraging rubrics to provide both fine-grained reward signals and offline guidance. We propose $\textbf{RGR-GRPO}$ (Reward and Guidance through Rubrics), a rubric-driven RL framework for multi-domain reasoning. RGR-GRPO enables LLMs to receive dense and informative rewards while exploring a larger solution space during GRPO training. Extensive experiments across 14 benchmarks spanning multiple domains demonstrate that RGR-GRPO consistently outperforms RL methods that rely solely on alternative reward schemes or offline guidance. Compared with verifiable online RL baseline, RGR-GRPO achieves average improvements of +7.0%, +5.4%, +8.4%, and +6.6% on mathematics, physics, chemistry, and general reasoning tasks, respectively. Notably, RGR-GRPO maintains stable entropy fluctuations during off-policy training and achieves superior pass@k performance, reflecting sustained exploration and effective breakthrough beyond existing performance bottlenecks.
Authors: Pattaraphon Kenny Wongchamcharoen, Paul Glasserman
Abstract: Large language models (LLMs) are increasingly used in finance and economics, where prompt-based attempts against look-ahead bias implicitly assume that models understand chronology. We test this fundamental question with a series of chronological ordering tasks with increasing complexities over facts the model already knows from pre-training. Our tasks cover (1) chronological ordering, (2) conditional sorting (filter, then order), and (3) anachronism detection. We evaluate GPT-4.1, Claude-3.7 Sonnet, with and without Extended Thinking (ET), and GPT-5 across multiple reasoning-effort settings. Across models, Exact match rate drops sharply as sequences lengthen even while rank correlations stay high as LLMs largely preserve local order but struggle to maintain a single globally consistent timeline. In conditional sorting, most failures stem from the filtering step rather than the ordering step, but GPT-5 and Claude-3.7 Sonnet with Extended Thinking outshine normal models significantly. Lastly, anachronism detection is found to be the easiest task for the LLMs but performance still declines with increasingly overlapping timelines or entities. Overall, our main contribution is showing that allocating explicit reasoning budget helps with chronological ordering with GPT-5 at medium/high reasoning effort achieving flawless ordering at all lengths and perfect conditional sorting (both self-filtered and given-subset), whereas low/minimal effort degrades with longer lists, mirroring earlier models. Our findings delineate limits of current LLMs on chronological tasks, providing insights into task complexity, and demonstrate scenarios in which reasoning helps. These patterns are important for the real-time application of LLMs in finance. We release all code and evaluation templates to support full reproducibility.
Authors: Alessio Pellegrino, Jacopo Mauro
Abstract: One of the long-standing goals in optimisation and constraint programming is to describe a problem in natural language and automatically obtain an executable, efficient model. Large language models appear to bring this vision closer, showing impressive results in automatically generating models for classical benchmarks. However, much of this apparent success may derive from data contamination rather than genuine reasoning: many standard CP problems are likely included in the training data of these models. To examine this hypothesis, we systematically rephrased and perturbed a set of well-known CSPLib problems to preserve their structure while modifying their context and introducing misleading elements. We then compared the models produced by three representative LLMs across original and modified descriptions. Our qualitative analysis shows that while LLMs can produce syntactically valid and semantically plausible models, their performance drops sharply under contextual and linguistic variation, revealing shallow understanding and sensitivity to wording.
Authors: Juan Miguel Lopez Alcaraz, Nils Strodthoff
Abstract: Despite the excelling performance of machine learning models, understanding their decisions remains a long-standing goal. Although commonly used attribution methods from explainable AI attempt to address this issue, they typically rely on associational rather than causal relationships. In this study, within the context of time series classification, we introduce a novel model-agnostic attribution method to assess the causal effect of concepts i.e., predefined segments within a time series, on classification outcomes. Our approach compares these causal attributions with closely related associational attributions, both theoretically and empirically. To estimate counterfactual outcomes, we use state-of-the-art diffusion models backed by state space models. We demonstrate the insights gained by our approach for a diverse set of qualitatively different time series classification tasks. Although causal and associational attributions might often share some similarities, in all cases they differ in important details, underscoring the risks associated with drawing causal conclusions from associational data alone. We believe that the proposed approach is also widely applicable in other domains to shed some light on the limits of associational attributions.
Authors: Pengfei Gu, Huimin Li, Yejia Zhang, Chaoli Wang, Danny Z. Chen
Abstract: Masked Autoencoders (MAEs) have been shown to be effective in pre-training Vision Transformers (ViTs) for natural and medical image analysis problems. By reconstructing missing pixel/voxel information in visible patches, a ViT encoder can aggregate contextual information for downstream tasks. But, existing MAE pre-training methods, which were specifically developed with the ViT architecture, lack the ability to capture geometric shape and spatial information, which is critical for medical image segmentation tasks. In this paper, we propose a novel extension of known MAEs for self pre-training (i.e., models pre-trained on the same target dataset) for 3D medical image segmentation. (1) We propose a new topological loss to preserve geometric shape information by computing topological signatures of both the input and reconstructed volumes, learning geometric shape information. (2) We introduce a pre-text task that predicts the positions of the centers and eight corners of 3D crops, enabling the MAE to aggregate spatial information. (3) We extend the MAE pre-training strategy to a hybrid state-of-the-art (SOTA) medical image segmentation architecture and co-pretrain it alongside the ViT. (4) We develop a fine-tuned model for downstream segmentation tasks by complementing the pre-trained ViT encoder with our pre-trained SOTA model. Extensive experiments on five public 3D segmentation datasets show the effectiveness of our new approach.
Authors: Faried Abu Zaid, Daniel Neider, Mustafa Yal\c{c}{\i}ner
Abstract: Formal verification has emerged as a promising method to ensure the safety and reliability of neural networks. However, many relevant properties, such as fairness or global robustness, pertain to the entire input space. If one applies verification techniques naively, the neural network is checked even on inputs that do not occur in the real world and have no meaning. To tackle this shortcoming, we propose the VeriFlow architecture as a flow-based density model tailored to allow any verification approach to restrict its search to some data distribution of interest. We argue that our architecture is particularly well suited for this purpose because of two major properties. First, we show that the transformation that is defined by our model is piecewise affine. Therefore, the model allows the usage of verifiers based on constraint solving with linear arithmetic. Second, upper density level sets (UDL) of the data distribution are definable via linear constraints in the latent space. As a consequence, representations of UDLs specified by a given probability are effectively computable in the latent space. This property allows for effective verification with a fine-grained, probabilistically interpretable control of how a-typical the inputs subject to verification are.
Authors: Francisco Valentini, Viviana Cotik, Dami\'an Furman, Ivan Bercovich, Edgar Altszyler, Juan Manuel P\'erez
Abstract: Information retrieval (IR) is the task of finding relevant documents in response to a user query. Although Spanish is the second most spoken native language, there are few Spanish IR datasets, which limits the development of information access tools for Spanish speakers. We introduce MessIRve, a large-scale Spanish IR dataset with almost 700,000 queries from Google's autocomplete API and relevant documents sourced from Wikipedia. MessIRve's queries reflect diverse Spanish-speaking regions, unlike other datasets that are translated from English or do not consider dialectal variations. The large size of the dataset allows it to cover a wide variety of topics, unlike smaller datasets. We provide a comprehensive description of the dataset, comparisons with existing datasets, and baseline evaluations of prominent IR models. Our contributions aim to advance Spanish IR research and improve information access for Spanish speakers.
Authors: Keyu Chen, Ziqian Bi, Qian Niu, Junyu Liu, Benji Peng, Sen Zhang, Ming Liu, Xinyuan Song, Zekun Jiang, Tianyang Wang, Ming Li, Xuanhe Pan, Jiawei Xu, Jinlang Wang, Pohsun Feng
Abstract: The application of TensorFlow pre-trained models in deep learning is explored, with an emphasis on practical guidance for tasks such as image classification and object detection. The study covers modern architectures, including ResNet, MobileNet, and EfficientNet, and demonstrates the effectiveness of transfer learning through real-world examples and experiments. A comparison of linear probing and model fine-tuning is presented, supplemented by visualizations using techniques like PCA, t-SNE, and UMAP, allowing for an intuitive understanding of the impact of these approaches. The work provides complete example code and step-by-step instructions, offering valuable insights for both beginners and advanced users. By integrating theoretical concepts with hands-on practice, the paper equips readers with the tools necessary to address deep learning challenges efficiently.
Authors: Tiantian Liu, Hongwei Yao, Feng Lin, Tong Wu, Zhan Qin, Kui Ren
Abstract: Embeddings have become a cornerstone in the functionality of large language models (LLMs) due to their ability to transform text data into rich, dense numerical representations that capture semantic and syntactic properties. These embedding vector databases serve as the long-term memory of LLMs, enabling efficient handling of a wide range of natural language processing tasks. However, the surge in popularity of embedding vector databases in LLMs has been accompanied by significant concerns about privacy leakage. Embedding vector databases are particularly vulnerable to embedding inversion attacks, where adversaries can exploit the embeddings to reverse-engineer and extract sensitive information from the original text data. Existing defense mechanisms have shown limitations, often struggling to balance security with the performance of downstream tasks. To address these challenges, we introduce Eguard, a novel defense mechanism designed to mitigate embedding inversion attacks. Eguard employs a transformer-based projection network and text mutual information optimization to safeguard embeddings while preserving the utility of LLMs. Our approach significantly reduces privacy risks, protecting over 95% of tokens from inversion while maintaining high performance across downstream tasks consistent with original embeddings.
Authors: Valery Istomin, Oleg Pereziabov, Ilya Afanasyev
Abstract: Camera-captured document images often suffer from geometric distortions caused by paper deformation, perspective distortion, and lens aberrations, significantly reducing OCR accuracy. This study develops an efficient automated method for document image dewarping that balances accuracy with computational efficiency. We propose a hybrid approach combining deep learning for document detection with classical computer vision for geometry restoration. YOLOv8 performs initial document segmentation and mask generation. Subsequently, classical CV techniques construct a topological 2D grid through cubic polynomial interpolation of document boundaries, followed by image remapping to correct nonlinear distortions. A new annotated dataset and open-source framework are provided to facilitate reproducibility and further research. Experimental evaluation against state-of-the-art methods (RectiNet, DocGeoNet, DocTr++) and mobile applications (DocScan, CamScanner, TapScanner) demonstrates superior performance. Our method achieves the lowest median Character Error Rate (CER=0.0235), Levenshtein Distance (LD=27.8), and highest Jaro--Winkler similarity (JW=0.902), approaching the quality of scanned originals. The approach requires significantly fewer computational resources and memory compared to pure deep learning solutions while delivering better OCR readability and geometry restoration quality. The proposed hybrid methodology effectively restores document geometry with computational efficiency superior to existing deep learning approaches, making it suitable for resource-constrained applications while maintaining high-quality document digitization. Project page: https://github.com/HorizonParadox/DRCCBI
Authors: Dimitrios Michail, Charalampos Davalas, Konstantinos Chafis, Lefki-Ioanna Panagiotou, Ioannis Prapas, Spyros Kondylatos, Nikolaos Ioannis Bountos, Ioannis Papoutsis
Abstract: With climate change intensifying fire weather conditions globally, accurate seasonal wildfire forecasting has become critical for disaster preparedness and ecosystem management. We introduce FireCastNet, a novel deep learning architecture that combines 3D convolutional encoding with GraphCast-based Graph Neural Networks (GNNs) to model complex spatio-temporal dependencies for global wildfire prediction. Our approach leverages the SeasFire dataset, a comprehensive multivariate Earth system datacube containing climate, vegetation, and human-related variables, to forecast burned area patterns up to six months in advance. FireCastNet treats the Earth as an interconnected graph, enabling it to capture both local fire dynamics and long-range teleconnections that influence wildfire behavior across different spatial and temporal scales. Through comprehensive benchmarking against state-of-the-art models including GRU, Conv-GRU, Conv-LSTM, U-TAE, and TeleViT, we demonstrate that FireCastNet achieves superior performance in global burned area forecasting, with particularly strong results in fire-prone regions such as Africa, South America, and Southeast Asia. Our analysis reveals that longer input time-series significantly improve prediction robustness, while spatial context integration enhances model performance across extended forecasting horizons. Additionally, we implement local area modeling techniques that provide enhanced spatial resolution and accuracy for region-specific predictions. These findings highlight the importance of modeling Earth system interactions for long-term wildfire prediction.
Authors: Jiangyong Yu, Changyong Shu, Sifan Zhou, Zichen Yu, Xing Hu, Yan Chen, Dawei Yang
Abstract: Camera-based multi-view 3D detection is crucial for autonomous driving. PETR and its variants (PETRs) excel in benchmarks but face deployment challenges due to high computational cost and memory footprint. Quantization is an effective technique for compressing deep neural networks by reducing the bit width of weights and activations. However, directly applying existing quantization methods to PETRs leads to severe accuracy degradation. This issue primarily arises from two key challenges: (1) significant magnitude disparity between multi-modal features-specifically, image features and camera-ray positional embeddings (PE), and (2) the inefficiency and approximation error of quantizing non-linear operators, which commonly rely on hardware-unfriendly computations. In this paper, we propose FQ-PETR, a fully quantized framework for PETRs, featuring three key innovations: (1) Quantization-Friendly LiDAR-ray Position Embedding (QFPE): Replacing multi-point sampling with LiDAR-prior-guided single-point sampling and anchor-based embedding eliminates problematic non-linearities (e.g., inverse-sigmoid) and aligns PE scale with image features, preserving accuracy. (2) Dual-Lookup Table (DULUT): This algorithm approximates complex non-linear functions using two cascaded linear LUTs, achieving high fidelity with minimal entries and no specialized hardware. (3) Quantization After Numerical Stabilization (QANS): Performing quantization after softmax numerical stabilization mitigates attention distortion from large inputs. On PETRs (e.g. PETR, StreamPETR, PETRv2, MV2d), FQ-PETR under W8A8 achieves near-floating-point accuracy (1% degradation) while reducing latency by up to 75%, significantly outperforming existing PTQ and QAT baselines.
Authors: Adib Karimi, Mohammad Mehdi Ebadzadeh
Abstract: We propose a novel Inverse Reinforcement Learning (IRL) method that mitigates the rigidity of fixed reward structures and the limited flexibility of implicit reward regularization. Building on the Maximum Entropy IRL framework, our approach incorporates a squared temporal-difference (TD) regularizer with adaptive targets that evolve dynamically during training, thereby imposing adaptive bounds on recovered rewards and promoting robust decision-making. To capture richer return information, we integrate distributional RL into the learning process. Empirically, our method achieves expert-level performance on complex MuJoCo and Adroit environments, surpassing baseline methods on the Humanoid-v2 task with limited expert demonstrations. Extensive experiments and ablation studies further validate the effectiveness of the approach and provide insights into reward dynamics in imitation learning. Our source code is available at https://github.com/adibka/RIZE.
Authors: Kyle Domico, Jean-Charles Noirot Ferrand, Ryan Sheatsley, Eric Pauley, Josiah Hanna, Patrick McDaniel
Abstract: Attacks on machine learning models have been extensively studied through stateless optimization. In this paper, we demonstrate how a reinforcement learning (RL) agent can learn a new class of attack algorithms that generate adversarial samples. Unlike traditional adversarial machine learning (AML) methods that craft adversarial samples independently, our RL-based approach retains and exploits past attack experience to improve the effectiveness and efficiency of future attacks. We formulate adversarial sample generation as a Markov Decision Process and evaluate RL's ability to (a) learn effective and efficient attack strategies and (b) compete with state-of-the-art AML. On two image classification benchmarks, our agent increases attack success rate by up to 13.2% and decreases the average number of victim model queries per attack by up to 16.9% from the start to the end of training. In a head-to-head comparison with state-of-the-art image attacks, our approach enables an adversary to generate adversarial samples with 17% more success on unseen inputs post-training. From a security perspective, this work demonstrates a powerful new attack vector that uses RL to train agents that attack ML models efficiently and at scale.
Authors: Changhe Chen, Xiaohao Xu, Xiangdong Wang, Xiaonan Huang
Abstract: Designing soft robots is a complex and iterative process that demands cross-disciplinary expertise in materials science, mechanics, and control, often relying on intuition and extensive experimentation. While foundation models, especially Large Language Models (LLMs), have demonstrated impressive reasoning abilities, their capacity to conduct embodied design remains largely unexplored. This paper introduces RoboCrafter-QA, a novel benchmark to evaluate whether LLMs can learn representations of soft robot designs that effectively bridge the gap between high-level task descriptions and low-level morphological and material choices. RoboCrafter-QA leverages the EvoGym simulator to generate a diverse set of soft robot design challenges, spanning robotic locomotion, manipulation, and balancing tasks. Our experiments with SOTA multi-modal LLMs reveal that while these models exhibit promising capabilities in learning design representations, they struggle with fine-grained distinctions between designs with subtle performance differences. To overcome these limitations, we finetune an efficient, open-source LLM that achieves SOTA performance on our benchmark, demonstrating superior capabilities in both design selection and direct generation of high-performing robot morphologies. Furthermore, we construct a physical replica of the modular soft robot and demonstrate a strong sim-to-real correlation, validating that superior benchmark performance has the potential to translate to effective real-world design selection. Our full system will be open-sourced to foster this exciting direction.
Authors: Miao Zhang, Sherif Abdulatif, Benedikt Loesch, Marco Altmann, Bin Yang
Abstract: Due to the significant effort required for data collection and annotation in 3D perception tasks, mixed sample data augmentation (MSDA) has been widely studied to generate diverse training samples by mixing existing data. Recently, many MSDA techniques have been developed for point clouds, but they mainly target LiDAR data, leaving their application to radar point clouds largely unexplored. In this paper, we examine the feasibility of applying existing MSDA methods to radar point clouds and identify several challenges in adapting these techniques. These obstacles stem from the radar's irregular angular distribution, deviations from a single-sensor polar layout in multi-radar setups, and point sparsity. To address these issues, we propose Class-Aware PillarMix (CAPMix), a novel MSDA approach that applies MixUp at the pillar level in 3D point clouds, guided by class labels. Unlike methods that rely a single mix ratio to the entire sample, CAPMix assigns an independent ratio to each pillar, boosting sample diversity. To account for the density of different classes, we use class-specific distributions: for dense objects (e.g., large vehicles), we skew ratios to favor points from another sample, while for sparse objects (e.g., pedestrians), we sample more points from the original. This class-aware mixing retains critical details and enriches each sample with new information, ultimately generating more diverse training data. Experimental results demonstrate that our method not only significantly boosts performance but also outperforms existing MSDA approaches across two datasets (Bosch Street and K-Radar). We believe that this straightforward yet effective approach will spark further investigation into MSDA techniques for radar data.
Authors: Yuwei Niu, Munan Ning, Mengren Zheng, Weiyang Jin, Bin Lin, Peng Jin, Jiaqi Liao, Chaoran Feng, Kunpeng Ning, Bin Zhu, Li Yuan
Abstract: Text-to-Image (T2I) models are capable of generating high-quality artistic creations and visual content. However, existing research and evaluation standards predominantly focus on image realism and shallow text-image alignment, lacking a comprehensive assessment of complex semantic understanding and world knowledge integration in text-to-image generation. To address this challenge, we propose \textbf{WISE}, the first benchmark specifically designed for \textbf{W}orld Knowledge-\textbf{I}nformed \textbf{S}emantic \textbf{E}valuation. WISE moves beyond simple word-pixel mapping by challenging models with 1000 meticulously crafted prompts across 25 subdomains in cultural common sense, spatio-temporal reasoning, and natural science. To overcome the limitations of traditional CLIP metric, we introduce \textbf{WiScore}, a novel quantitative metric for assessing knowledge-image alignment. Through comprehensive testing of 20 models (10 dedicated T2I models and 10 unified multimodal models) using 1,000 structured prompts spanning 25 subdomains, our findings reveal significant limitations in their ability to effectively integrate and apply world knowledge during image generation, highlighting critical pathways for enhancing knowledge incorporation and application in next-generation T2I models. Code and data are available at \href{https://github.com/PKU-YuanGroup/WISE}{PKU-YuanGroup/WISE}.
Authors: Jian Zhu, Zhengyu Jia, Tian Gao, Jiaxin Deng, Shidi Li, Lang Zhang, Fu Liu, Peng Jia, Xianpeng Lang
Abstract: Advanced end-to-end autonomous driving systems predict other vehicles' motions and plan ego vehicle's trajectory. The world model that can foresee the outcome of the trajectory has been used to evaluate the autonomous driving system. However, existing world models predominantly emphasize the trajectory of the ego vehicle and leave other vehicles uncontrollable. This limitation hinders their ability to realistically simulate the interaction between the ego vehicle and the driving scenario. In this paper, we propose a driving World Model named EOT-WM, unifying Ego-Other vehicle Trajectories in videos for driving simulation. Specifically, it remains a challenge to match multiple trajectories in the BEV space with each vehicle in the video to control the video generation. We first project ego-other vehicle trajectories in the BEV space into the image coordinate for vehicle-trajectory match via pixel positions. Then, trajectory videos are encoded by the Spatial-Temporal Variational Auto Encoder to align with driving video latents spatially and temporally in the unified visual space. A trajectory-injected diffusion Transformer is further designed to denoise the noisy video latents for video generation with the guidance of ego-other vehicle trajectories. In addition, we propose a metric based on control latent similarity to evaluate the controllability of trajectories. Extensive experiments are conducted on the nuScenes dataset, and the proposed model outperforms the state-of-the-art method by 30% in FID and 55% in FVD. The model can also predict unseen driving scenes with self-produced trajectories.
Authors: Shubham Kumar, Narendra Ahuja
Abstract: Deep vision models perform input-output computations that are hard to interpret. Concept-based explanation methods (CBEMs) increase interpretability by re-expressing parts of the model with human-understandable semantic units, or concepts. Checking if the derived explanations are faithful -- that is, they represent the model's internal computation -- requires a surrogate that combines concepts to compute the output. Simplifications made for interpretability inevitably reduce faithfulness, resulting in a tradeoff between the two. State-of-the-art unsupervised CBEMs (U-CBEMs) have reported increasingly interpretable concepts, while also being more faithful to the model. However, we observe that the reported improvement in faithfulness artificially results from either (1) using overly complex surrogates, which introduces an unmeasured cost to the explanation's interpretability, or (2) relying on deletion-based approaches that, as we demonstrate, do not properly measure faithfulness. We propose Surrogate Faithfulness (SURF), which (1) replaces prior complex surrogates with a simple, linear surrogate that measures faithfulness without changing the explanation's interpretability and (2) introduces well-motivated metrics that assess loss across all output classes, not just the predicted class. We validate SURF with a measure-over-measure study by proposing a simple sanity check -- explanations with random concepts should be less faithful -- which prior surrogates fail. SURF enables the first reliable faithfulness benchmark of U-CBEMs, revealing that many visually compelling U-CBEMs are not faithful. Code to be released.
Authors: Md Asaduzzaman Jabin, Hanqi Jiang, Yiwei Li, Patrick Kaggwa, Eugene Douglass, Juliet N. Sekandi, Tianming Liu
Abstract: Chronic diseases, including diabetes, hypertension, asthma, HIV-AIDS, epilepsy, and tuberculosis, necessitate rigorous adherence to medication to avert disease progression, manage symptoms, and decrease mortality rates. Adherence is frequently undermined by factors including patient behavior, caregiver support, elevated medical costs, and insufficient healthcare infrastructure. We propose AdCare-VLM, a specialized LLaVA-based multimodal large vision language model (LVLM) by introducing a unified visual latent space with pre-alignment to facilitate visual question answering (VQA) concerning medication adherence through patient videos. We employ a private dataset comprising 806 custom-annotated tuberculosis (TB) medication monitoring videos, which have been labeled by clinical experts, to fine-tune the model for adherence pattern detection. We present LLM-TB-VQA, a detailed medical adherence VQA dataset that encompasses positive, negative, and ambiguous adherence cases. Our method identifies correlations between visual features, such as the clear visibility of the patient's face, medication, water intake, and the act of ingestion, and their associated medical concepts in captions. This facilitates the integration of aligned visual-linguistic representations and improves multimodal interactions. Experimental results indicate that our method surpasses parameter-efficient fine-tuning (PEFT) enabled VLM models, such as LLaVA-V1.5 and Chat-UniVi, with absolute improvements ranging from 3.1% to 3.54% across pre-trained, regular, and low-rank adaptation (LoRA) configurations. Comprehensive ablation studies and attention map visualizations substantiate our approach, enhancing interpretability.
Authors: Nikolaos Louloudakis, Ajitha Rajan
Abstract: With over 760 stars on GitHub and being part of the official ONNX repository, the ONNX Optimizer is the default tool for applying graph-based optimizations to ONNX models. Despite its widespread use, its ability to maintain model accuracy during optimization has not been thoroughly investigated. In this work, we present OODTE, a utility designed to automatically and comprehensively evaluate the correctness of the ONNX Optimizer. OODTE adopts a straightforward yet powerful differential testing and evaluation methodology, which can be readily adapted for use with other compiler optimizers. Specifically, OODTE takes a collection of ONNX models, applies optimizations, and executes both the original and optimized versions across a user-defined input set, automatically capturing any issues encountered during optimization. When discrepancies in accuracy arise, OODTE iteratively isolates the responsible optimization pass by repeating the process at a finer granularity. We applied OODTE to 130 well-known models from the official ONNX Model Hub, spanning diverse tasks including classification, object detection, semantic segmentation, text summarization, question answering, and sentiment analysis. Our evaluation revealed that 9.2% of the model instances either caused the optimizer to crash or led to the generation of invalid models using default optimization strategies. Additionally, 30% of classification models and 16.6% of object detection and segmentation models exhibited differing outputs across original and optimized versions, whereas models focused on text-related tasks were generally robust to optimization. OODTE uncovered 15 issues-14 previously unknown-affecting 9 of 47 optimization passes and the optimizer overall. All issues were reported to the ONNX Optimizer team. OODTE offers a simple but effective framework for validating AI model optimizers, applicable beyond the ONNX ecosystem.
Authors: Steven Bedrick, A. Seza Do\u{g}ru\"oz, Sergiu Nisioi
Abstract: Synthetic data sets are used across linguistic domains and NLP tasks, particularly in scenarios where authentic data is limited (or even non-existent). One such domain is that of clinical (healthcare) contexts, where there exist significant and long-standing challenges (e.g., privacy, anonymization, and data governance) which have led to the development of an increasing number of synthetic datasets. One increasingly important category of clinical dataset is that of clinical dialogues which are especially sensitive and difficult to collect, and as such are commonly synthesized. While such synthetic datasets have been shown to be sufficient in some situations, little theory exists to inform how they may be best used and generalized to new applications. In this paper, we provide an overview of how synthetic datasets are created, evaluated and being used for dialogue related tasks in the medical domain. Additionally, we propose a novel typology for use in classifying types and degrees of data synthesis, to facilitate comparison and evaluation.
Authors: Amir Rezaei Balef, Claire Vernade, Katharina Eggensperger
Abstract: The Combined Algorithm Selection and Hyperparameter optimization (CASH) is a challenging resource allocation problem in the field of AutoML. We propose MaxUCB, a max k-armed bandit method to trade off exploring different model classes and conducting hyperparameter optimization. MaxUCB is specifically designed for the light-tailed and bounded reward distributions arising in this setting and, thus, provides an efficient alternative compared to classic max k-armed bandit methods assuming heavy-tailed reward distributions. We theoretically and empirically evaluate our method on four standard AutoML benchmarks, demonstrating superior performance over prior approaches. We make our code and data available at https://github.com/amirbalef/CASH_with_Bandits
Authors: Hao Wu, Yuan Gao, Chang Liu, Fan Xu, Fan Zhang, Zhihong Zhu, Yuqi Li, Xian Wu, Yuxuan Liang, Li Liu, Qingsong Wen, Kun Wang, Yu Zheng, Xiaomeng Huang
Abstract: Accurately predicting the long-term evolution of turbulence is crucial for advancing scientific understanding and optimizing engineering applications. However, existing deep learning methods face significant bottlenecks in long-term autoregressive prediction, which exhibit excessive smoothing and fail to accurately track complex fluid dynamics. Our extensive experimental and spectral analysis of prevailing methods provides an interpretable explanation for this shortcoming, identifying Spectral Bias as the core obstacle. Concretely, spectral bias is the inherent tendency of models to favor low-frequency, smooth features while overlooking critical high-frequency details during training, thus reducing fidelity and causing physical distortions in long-term predictions. Building on this insight, we propose Turb-L1, an innovative turbulence prediction method, which utilizes a Hierarchical Dynamics Synthesis mechanism within a multi-grid architecture to explicitly overcome spectral bias. It accurately captures cross-scale interactions and preserves the fidelity of high-frequency dynamics, enabling reliable long-term tracking of turbulence evolution. Extensive experiments on the 2D turbulence benchmark show that Turb-L1 demonstrates excellent performance: (I) In long-term predictions, it reduces Mean Squared Error (MSE) by $80.3\%$ and increases Structural Similarity (SSIM) by over $9\times$ compared to the SOTA baseline, significantly improving prediction fidelity. (II) It effectively overcomes spectral bias, accurately reproducing the full enstrophy spectrum and maintaining physical realism in high-wavenumber regions, thus avoiding the spectral distortions or spurious energy accumulation seen in other methods.
Authors: Patrick Yubeaton, Andre Nakkab, Weihua Xiao, Luca Collini, Ramesh Karri, Chinmay Hegde, Siddharth Garg
Abstract: This paper introduces VeriThoughts, a novel dataset designed for reasoning-based Verilog code generation. We establish a new benchmark framework grounded in formal verification methods to evaluate the quality and correctness of generated hardware descriptions. Additionally, we present a suite of specialized small-scale models optimized specifically for Verilog generation. Our work addresses the growing need for automated hardware design tools that can produce verifiably correct implementations from high-level specifications, potentially accelerating the hardware development process while maintaining rigorous correctness guarantees. Our code and data are available at \href{https://github.com/wilyub/VeriThoughts}{this URL}.
Authors: Tongyu Lu, Charlotta-Marlena Geist, Jan Melechovsky, Abhinaba Roy, Dorien Herremans
Abstract: We propose MelodySim, a melody-aware music similarity model and dataset for plagiarism detection. First, we introduce a novel method to construct a dataset focused on melodic similarity. By augmenting Slakh2100, an existing MIDI dataset, we generate variations of each piece while preserving the melody through modifications such as note splitting, arpeggiation, minor track dropout, and re-instrumentation. A user study confirms that positive pairs indeed contain similar melodies, while other musical tracks are significantly changed. Second, we develop a segment-wise melodic-similarity detection model that uses a MERT encoder and applies a triplet neural network to capture melodic similarity. The resulting decision matrix highlights where plagiarism might occur. The experiments show that our model is able to outperform baseline models in detecting similar melodic fragments on the MelodySim test set.
Authors: Rajat Rasal, Avinash Kori, Ben Glocker
Abstract: Identifiable causal representation learning seeks to uncover the true causal relationships underlying a data generation process. In medical imaging, this presents opportunities to improve the generalisability and robustness of task-specific latent features. This work introduces the concept of grouping observations to learn identifiable representations for disease classification in chest X-rays via an end-to-end framework. Our experiments demonstrate that these causal representations improve generalisability and robustness across multiple classification tasks when grouping is used to enforce invariance w.r.t race, sex, and imaging views.
Authors: Soumik Dey, Benjamin Braun, Naveen Ravipati, Hansi Wu, Binbin Li
Abstract: E-commerce sellers are advised to bid on keyphrases to boost their advertising campaigns. These keyphrases must be relevant to prevent irrelevant items from cluttering search systems and to maintain positive seller perception. It is vital that keyphrase suggestions align with seller, search and buyer judgments. Given the challenges in collecting negative feedback in these systems, LLMs have been used as a scalable proxy to human judgments. This paper presents an empirical study on a major ecommerce platform of a distillation framework involving an LLM teacher, a cross-encoder assistant and a bi-encoder Embedding Based Retrieval (EBR) student model, aimed at mitigating click-induced biases in keyphrase recommendations.
Authors: Ferran Agullo, Joan Oliveras, Chen Wang, Alberto Gutierrez-Torre, Olivier Tardieu, Alaa Youssef, Jordi Torres, Josep Ll. Berral
Abstract: With the rapid adoption of Large Language Models (LLMs), LLM-adapters have become increasingly common, providing lightweight specialization of large-scale models. Serving hundreds or thousands of these adapters on a single GPU allows request aggregation, increasing throughput, but may also cause request starvation if GPU memory limits are exceeded. To address this issue, this study focuses on determining the joint configuration of concurrent and parallel adapters that maximizes GPU throughput without inducing starvation, given heterogeneous adapter and traffic properties. We propose a data-driven ML approach leveraging interpretable models to tackle this caching problem and introduce the first Digital Twin capable of reproducing an LLM-adapter serving system, enabling efficient training data generation. Experiments with the vLLM framework and LoRA adapters show that the Digital Twin reproduces throughput within 5.1% of real results, while the ML approach predicts optimal numbers of concurrent and parallel adapters with an error of at most 7.2% under heterogeneous, real-world workloads. The code is publicly available at https://github.com/FerranAgulloLopez/GPULLMAdapterOptimization.
URLs: https://github.com/FerranAgulloLopez/GPULLMAdapterOptimization.
Authors: Daoze Zhang, Chenghan Fu, Zhanheng Nie, Jianyu Liu, Wanxian Guan, Yuan Gao, Jun Song, Pengjie Wang, Jian Xu, Bo Zheng
Abstract: With the rapid advancement of e-commerce, exploring general representations rather than task-specific ones has attracted increasing research attention. For product understanding, although existing discriminative dual-flow architectures drive progress in this field, they inherently struggle to model the many-to-one alignment between multiple images and texts of products. Therefore, we argue that generative Multimodal Large Language Models (MLLMs) hold significant potential for improving product representation learning. Nevertheless, achieving this goal still remains non-trivial due to several key challenges: the lack of multimodal and aspect-aware modeling modules in typical LLMs; the common presence of background noise in product images; and the absence of a standard benchmark for evaluation. To address these issues, we propose the first generative MLLM-based model named MOON for product representation learning. Our method (1) employs a guided Mixture-of-Experts (MoE) module for targeted modeling of multimodal and aspect-specific product content; (2) effectively detects core semantic regions in product images to mitigate the distraction and interference caused by background noise; and (3) introduces the specialized negative sampling strategy to increase the difficulty and diversity of negative samples. In addition, we release a large-scale multimodal benchmark MBE for various product understanding tasks. Experimentally, our model demonstrates competitive zero-shot performance on both our benchmark and the public dataset, showcasing strong generalization across various downstream tasks, including cross-modal retrieval, product classification, and attribute prediction. Furthermore, the case study and visualization illustrate the effectiveness of MOON for product understanding.
Authors: Zhendong Yang, Jie Wang, Liansong Zong, Xiaorong Liu, Quan Qian, Shiqian Chen
Abstract: Few-Shot Class-Incremental Fault Diagnosis (FSC-FD), which aims to continuously learn from new fault classes with only a few samples without forgetting old ones, is critical for real-world industrial systems. However, this challenging task severely amplifies the issues of catastrophic forgetting of old knowledge and overfitting on scarce new data. To address these challenges, this paper proposes a novel framework built upon Dual-Granularity Representations, termed the Dual-Granularity Guidance Network (DGGN). Our DGGN explicitly decouples feature learning into two parallel streams: 1) a fine-grained representation stream, which utilizes a novel Multi-Order Interaction Aggregation module to capture discriminative, class-specific features from the limited new samples. 2) a coarse-grained representation stream, designed to model and preserve general, class-agnostic knowledge shared across all fault types. These two representations are dynamically fused by a multi-semantic cross-attention mechanism, where the stable coarse-grained knowledge guides the learning of fine-grained features, preventing overfitting and alleviating feature conflicts. To further mitigate catastrophic forgetting, we design a Boundary-Aware Exemplar Prioritization strategy. Moreover, a decoupled Balanced Random Forest classifier is employed to counter the decision boundary bias caused by data imbalance. Extensive experiments on the TEP benchmark and a real-world MFF dataset demonstrate that our proposed DGGN achieves superior diagnostic performance and stability compared to state-of-the-art FSC-FD approaches. Our code is publicly available at https://github.com/MentaY/DGGN
Authors: Riccardo Pozzi, Matteo Palmonari, Andrea Coletta, Luigi Bellomarini, Jens Lehmann, Sahar Vahdati
Abstract: Knowledge gaps and hallucinations are persistent challenges for Large Language Models (LLMs), which generate unreliable responses when lacking the necessary information to fulfill user instructions. Existing approaches, such as Retrieval-Augmented Generation (RAG) and tool use, aim to address these issues by incorporating external knowledge. Yet, they rely on additional models or services, resulting in complex pipelines, potential error propagation, and often requiring the model to process a large number of tokens. In this paper, we present a scalable method that enables LLMs to access external knowledge without depending on retrievers or auxiliary models. Our approach uses constrained generation with a pre-built prefix-tree index. Triples from a Knowledge Graph are verbalized in textual facts, tokenized, and indexed in a prefix tree for efficient access. During inference, to acquire external knowledge, the LLM generates facts with constrained generation which allows only sequences of tokens that form an existing fact. We evaluate our proposal on Question Answering and show that it scales to large knowledge bases (800 million facts), adapts to domain-specific data, and achieves effective results. These gains come with minimal generation-time overhead. ReFactX code is available at https://github.com/rpo19/ReFactX.
Authors: Alex Cuellar, Ho Chit Siu, Julie A Shah
Abstract: As robots' manipulation capabilities improve for pick-and-place tasks (e.g., object packing, sorting, and kitting), methods focused on understanding human-acceptable object configurations remain limited expressively with regard to capturing spatial relationships important to humans. To advance robotic understanding of human rules for object arrangement, we introduce positionally-augmented RCC (PARCC), a formal logic framework based on region connection calculus (RCC) for describing the relative position of objects in space. Additionally, we introduce an inference algorithm for learning PARCC specifications via demonstrations. Finally, we present the results from a human study, which demonstrate our framework's ability to capture a human's intended specification and the benefits of learning from demonstration approaches over human-provided specifications.
Authors: Tongtong Feng, Xin Wang, Feilin Han, Leping Zhang, Wenwu Zhu
Abstract: Swarm UAV autonomous flight for Embodied Long-Horizon (ELH) tasks is crucial for advancing the low-altitude economy. However, existing methods focus only on specific basic tasks due to dataset limitations, failing in real-world deployment for ELH tasks. ELH tasks are not mere concatenations of basic tasks, requiring handling long-term dependencies, maintaining embodied persistent states, and adapting to dynamic goal shifts. This paper presents U2UData+, the first large-scale swarm UAV autonomous flight dataset for ELH tasks and the first scalable swarm UAV data online collection and algorithm closed-loop verification platform. The dataset is captured by 15 UAVs in autonomous collaborative flights for ELH tasks, comprising 12 scenes, 720 traces, 120 hours, 600 seconds per trajectory, 4.32M LiDAR frames, and 12.96M RGB frames. This dataset also includes brightness, temperature, humidity, smoke, and airflow values covering all flight routes. The platform supports the customization of simulators, UAVs, sensors, flight algorithms, formation modes, and ELH tasks. Through a visual control window, this platform allows users to collect customized datasets through one-click deployment online and to verify algorithms by closed-loop simulation. U2UData+ also introduces an ELH task for wildlife conservation and provides comprehensive benchmarks with 9 SOTA models. U2UData+ can be found at https://fengtt42.github.io/U2UData-2/.
Authors: Seungkyu Lee, Nalim Kim, Yohan Jo
Abstract: Tool agents -- LLM-based systems that interact with external APIs -- offer a way to execute real-world tasks. However, as tasks become increasingly complex, these agents struggle to identify and call the correct APIs in the proper order. To tackle this problem, we investigate converting API documentation into a structured API graph that captures API dependencies and leveraging it for multi-tool queries that require compositional API calls. To support this, we introduce In-N-Out, the first expert-annotated dataset of API graphs built from two real-world API benchmarks and their documentation. Using In-N-Out significantly improves performance on both tool retrieval and multi-tool query generation, nearly doubling that of LLMs using documentation alone. Moreover, graphs generated by models fine-tuned on In-N-Out close 90% of this gap, showing that our dataset helps models learn to comprehend API documentation and parameter relationships. Our findings highlight the promise of using explicit API graphs for tool agents and the utility of In-N-Out as a valuable resource. We will release the dataset and code publicly.
Authors: Ibne Farabi Shihab, Sanjeda Akter, Anuj Sharma
Abstract: We introduce the first differentiable approximation of range-partition entropy, a complexity measure from computational geometry that directly bounds algorithmic runtime. Unlike architectural modifications, our method is a complementary regularizer that provides orthogonal efficiency gains when combined with existing optimizations. We establish theoretical guarantees in computational geometry, achieving 4--5$\times$ provable speedups on convex hull and triangulation with $<$0.2\% error. On ImageNet-1K with ViT-Base, entropy regularization achieves 80.1\% top-1 accuracy at 80\% sparsity (1.60$\times$ standalone speedup), and when combined with FlashAttention yields 2.07$\times$ speedup versus 1.63$\times$ for FlashAttention alone. On large language models (LLaMA-2 7B, Mistral-7B, Phi-2), we achieve 1.48--1.60$\times$ inference speedups at 70--75\% sparsity with minimal quality degradation (ROUGE-L drops of 0.3--0.4 points, perplexity increase of 0.9). Unlike prior regularization methods that target output distributions, we directly minimize representation complexity, yielding both efficiency gains and improved robustness through semantically structured sparsity patterns (IoU 0.73 vs 0.41 for magnitude pruning, CIFAR-100-C mCE 48.7 vs 55.4). Benefits are strongest for geometry and vision transformers, with more modest but measurable gains on LLMs, demonstrating that complexity regularization offers a principled pathway to joint efficiency-robustness optimization.
Authors: Mohammad Rashed Albous, Abdel Latef Anouze
Abstract: Artificial intelligence (AI) is rapidly transforming public-sector processes worldwide, yet standardized measures rarely address the unique drivers, governance models, and cultural nuances of the Gulf Cooperation Council (GCC) countries. This study employs a theory-driven foundation derived from an in-depth analysis of literature review and six National AI Strategies (NASs), coupled with a data-driven approach that utilizes a survey of 203 mid- and senior-level government employees and advanced statistical techniques (K-Means clustering, Principal Component Analysis, and Partial Least Squares Structural Equation Modeling). By combining policy insights with empirical evidence, the research develops and validates a novel AI Adoption Index specifically tailored to the GCC public sector. Findings indicate that robust technical infrastructure and clear policy mandates exert the strongest influence on successful AI implementations, overshadowing organizational readiness in early adoption stages. The combined model explains 70% of the variance in AI outcomes, suggesting that resource-rich environments and top-down policy directives can drive rapid but uneven technology uptake. By consolidating key dimensions (Technical Infrastructure (TI), Organizational Readiness (OR), and Governance Environment (GE)) into a single composite index, this study provides a holistic yet context-sensitive tool for benchmarking AI maturity. The index offers actionable guidance for policymakers seeking to harmonize large-scale deployments with ethical and regulatory standards. Beyond advancing academic discourse, these insights inform more strategic allocation of resources, cross-country cooperation, and capacity-building initiatives, thereby supporting sustained AI-driven transformation in the GCC region and beyond.
Authors: Mahmudul Islam Masum, Miad Islam
Abstract: As local AI grows in popularity, there is a critical gap between the benchmark performance of object detectors and their practical viability on consumer-grade hardware. While models like YOLOv10s promise real-time speeds, these metrics are typically achieved on high-power, desktop-class GPUs. This paper reveals that on resource-constrained systems, such as laptops with RTX 4060 GPUs, performance is not compute-bound but is instead dominated by system-level bottlenecks, as illustrated by a simple bottleneck test. To overcome this hardware-level constraint, we introduce a Two-Pass Adaptive Inference algorithm, a model-independent approach that requires no architectural changes. This study mainly focuses on adaptive inference strategies and undertakes a comparative analysis of architectural early-exit and resolution-adaptive routing, highlighting their respective trade-offs within a unified evaluation framework. The system uses a fast, low-resolution pass and only escalates to a high-resolution model pass when detection confidence is low. On a 5000-image COCO dataset, our method achieves a 1.85x speedup over a PyTorch Early-Exit baseline, with a modest mAP loss of 5.51%. This work provides a practical and reproducible blueprint for deploying high-performance, real-time AI on consumer-grade devices by shifting the focus from pure model optimization to hardware-aware inference strategies that maximize throughput.
Authors: Advait Gosai, Arun Kavishwar, Stephanie L. McNamara, Soujanya Samineni, Renato Umeton, Alexander Chowdhury, William Lotter
Abstract: Recent work has shown promising performance of frontier large language models (LLMs) and their multimodal counterparts in medical quizzes and diagnostic tasks, highlighting their potential for broad clinical utility given their accessible, general-purpose nature. However, beyond diagnosis, a fundamental aspect of medical image interpretation is the ability to localize pathological findings. Evaluating localization not only has clinical and educational relevance but also provides insight into a model's spatial understanding of anatomy and disease. Here, we systematically assess two general-purpose MLLMs (GPT-4 and GPT-5) and a domain-specific model (MedGemma) in their ability to localize pathologies on chest radiographs, using a prompting pipeline that overlays a spatial grid and elicits coordinate-based predictions. Averaged across nine pathologies in the CheXlocalize dataset, GPT-5 exhibited a localization accuracy of 49.7%, followed by GPT-4 (39.1%) and MedGemma (17.7%), all lower than a task-specific CNN baseline (59.9%) and a radiologist benchmark (80.1%). Despite modest performance, error analysis revealed that GPT-5's predictions were largely in anatomically plausible regions, just not always precisely localized. GPT-4 performed well on pathologies with fixed anatomical locations, but struggled with spatially variable findings and exhibited anatomically implausible predictions more frequently. MedGemma demonstrated the lowest performance on all pathologies, but showed improvements when provided examples through few shot prompting. Our findings highlight both the promise and limitations of current MLLMs in medical imaging and underscore the importance of integrating them with task-specific tools for reliable use.
Authors: Longxuan Yu, Xing Shi, Xianghao Kong, Tong Jia, Greg Ver Steeg
Abstract: Mutual information (MI) is one of the most general ways to measure relationships between random variables, but estimating this quantity for complex systems is challenging. Denoising diffusion models have recently set a new bar for density estimation, so it is natural to consider whether these methods could also be used to improve MI estimation. Using the recently introduced information-theoretic formulation of denoising diffusion models, we show the diffusion models can be used in a straightforward way to estimate MI. In particular, the MI corresponds to half the gap in the Minimum Mean Square Error (MMSE) between conditional and unconditional diffusion, integrated over all Signal-to-Noise-Ratios (SNRs) in the noising process. Our approach not only passes self-consistency tests but also outperforms traditional and score-based diffusion MI estimators. Furthermore, our method leverages adaptive importance sampling to achieve scalable MI estimation, while maintaining strong performance even when the MI is high.
Authors: Sameep Chattopadhyay, Nikhil Karamchandani, Sharayu Moharir
Abstract: Online learning to rank (OLTR) plays a critical role in information retrieval and machine learning systems, with a wide range of applications in search engines and content recommenders. However, despite their extensive adoption, the susceptibility of OLTR algorithms to coordinated adversarial attacks remains poorly understood. In this work, we present a novel framework for attacking some of the widely used OLTR algorithms. Our framework is designed to promote a set of target items so that they appear in the list of top-K recommendations for T - o(T) rounds, while simultaneously inducing linear regret in the learning algorithm. We propose two novel attack strategies: CascadeOFA for CascadeUCB1 and PBMOFA for PBM-UCB . We provide theoretical guarantees showing that both strategies require only O(log T) manipulations to succeed. Additionally, we supplement our theoretical analysis with empirical results on real-world data.
Authors: Jintao Zhang, Haoxu Wang, Kai Jiang, Shuo Yang, Kaiwen Zheng, Haocheng Xi, Ziteng Wang, Hongzhou Zhu, Min Zhao, Ion Stoica, Joseph E. Gonzalez, Jun Zhu, Jianfei Chen
Abstract: In Diffusion Transformer (DiT) models, particularly for video generation, attention latency is a major bottleneck due to the long sequence length and the quadratic complexity. We find that attention weights can be separated into two parts: a small fraction of large weights with high rank and the remaining weights with very low rank. This naturally suggests applying sparse acceleration to the first part and low-rank acceleration to the second. Based on this finding, we propose SLA (Sparse-Linear Attention), a trainable attention method that fuses sparse and linear attention to accelerate diffusion models. SLA classifies attention weights into critical, marginal, and negligible categories, applying O(N^2) attention to critical weights, O(N) attention to marginal weights, and skipping negligible ones. SLA combines these computations into a single GPU kernel and supports both forward and backward passes. With only a few fine-tuning steps using SLA, DiT models achieve a 20x reduction in attention computation, resulting in significant acceleration without loss of generation quality. Experiments show that SLA reduces attention computation by 95% without degrading end-to-end generation quality, outperforming baseline methods. In addition, we implement an efficient GPU kernel for SLA, which yields a 13.7x speedup in attention computation and a 2.2x end-to-end speedup in video generation on Wan2.1-1.3B. The code is available at https://github.com/thu-ml/SLA.
Authors: Shijie Lian, Changti Wu, Laurence Tianruo Yang, Hang Yuan, Bin Yu, Lei Zhang, Kai Chen
Abstract: Spatial intelligence spans a rich suite of abilities, including visualising and transforming shapes, mentally rotating objects, judging relational positions and containment, and estimating numerosity. However, it still remains a critical unresolved challenge for Multimodal Large Language Models (MLLMs). To fill this gap, we propose to treat Euclidean geometry problem-solving as a surrogate task. Specifically, we meticulously constructed a curated multimodal dataset, called Euclid30K, comprising approximately 30K plane and solid geometry problems. Furthermore, to enable the model to learn and apply Euclidean principles from these geometry problems, we fine-tuned seven model variants (spanning 3--72B parameters) from the Qwen2.5VL, Qwen3VL, and RoboBrain2.0 families using Group Relative Policy Optimization (GRPO), inspiring the models to identify shapes, count, and relate entities, and perform multi-step deductive reasoning using Euclidean principles. Our experiments demonstrate that the resulting models achieve substantial zero-shot gains across four spatial reasoning benchmarks (Super-CLEVR, Omni3DBench, VSI-Bench, and MindCube) without any task-specific adaptations. Notably, after training on the Euclid30K, the mean VSI-Bench accuracy rose from 36.6\% to 41.8\% (+5.2\%), and the mean MindCube accuracy rose from 31.4\% to 38.1\% (+6.7\%). To our knowledge, this is the first systematic study showing that geometry-centric fine-tuning can confer vision-language models with broadly transferable spatial skills. Code and Euclid30K dataset can be found in \href{https://zgca-ai4edu.github.io/Euclids_Gift}{this}.
Authors: Guoliang Gong, Man Yu
Abstract: Computed Tomography (CT) is a vital diagnostic tool in clinical practice, yet the health risks associated with ionizing radiation cannot be overlooked. Low-dose CT (LDCT) helps mitigate radiation exposure but simultaneously leads to reduced image quality. Consequently, researchers have sought to reconstruct clear images from LDCT scans using artificial intelligence-based image enhancement techniques. However, these studies typically rely on synthetic LDCT images for algorithm training, which introduces significant domain-shift issues and limits the practical effectiveness of these algorithms in real-world scenarios. To address this challenge, we constructed a real-world paired lung dataset, referred to as Patient-uLDCT (ultra-low-dose CT), by performing multiple scans on volunteers. The radiation dose for the low-dose images in this dataset is only 2% of the normal dose, substantially lower than the conventional 25% low-dose and 10% ultra-low-dose levels. Furthermore, to resolve the anatomical misalignment between normal-dose and uLDCT images caused by respiratory motion during acquisition, we propose a novel purification strategy to construct corresponding aligned image pairs. Finally, we introduce a Frequency-domain Flow Matching model (FFM) that achieves excellent image reconstruction performance. Code is available at https://github.com/MonkeyDadLufy/flow-matching.
Authors: Daniel Jason Tan, Jiayang Chen, Dilruk Perera, Kay Choong See, Mengling Feng
Abstract: ICU enteral feeding remains sub-optimal due to limited personalization and uncertainty about appropriate calorie, protein, and fluid targets, particularly under rapidly changing metabolic demands and heterogeneous patient responses. This study introduces DeepEN, a reinforcement learning (RL)-based framework that personalizes enteral nutrition (EN) dosing for critically ill patients using electronic health record data. DeepEN was trained on over 11,000 ICU patients from the MIMIC-IV database to generate 4-hourly, patient-specific targets for caloric, protein, and fluid intake. The model's state space integrates demographics, comorbidities, vital signs, laboratory results, and prior interventions relevant to nutritional management, while its reward function balances short-term physiological and nutrition-related goals with long-term survival. A dueling double deep Q-network with Conservative Q-Learning regularization is used to ensure safe and reliable policy learning from retrospective data. DeepEN achieved a 3.7 $\pm$ 0.17 percentage-point absolute reduction in estimated mortality compared with the clinician policy (18.8% vs 22.5%) and higher expected returns compared with guideline-based dosing (11.89 vs 8.11), with improvements in key nutritional biomarkers. U-shaped associations between deviations from clinician dosing and mortality suggest that the learned policy aligns with high-value clinician actions while diverging from suboptimal ones. These findings demonstrate the feasibility of conservative offline RL for individualized EN therapy and suggest that data-driven personalization may improve outcomes beyond guideline- or heuristic-based approaches.
Authors: Ha Min Son, Huan Ren, Xin Liu, Zhe Zhao
Abstract: Android is the largest mobile platform, yet automatically building applications remains a practical challenge. While Large Language Models (LLMs) show promise for code repair, their use for fixing Android build errors remains underexplored. To address this gap, we first introduce AndroidBuildBench, a benchmark of 1,019 build failures curated from the commit histories of 43 open-source Android projects. Each problem is paired with a verified solution from a subsequent commit, ensuring that fixes are feasible. Second, we propose GradleFixer, an LLM agent with domain-specific tools for inspecting and manipulating the Gradle build environment. GradleFixer achieves a resolve rate of 81.4% (pass@1), significantly outperforming a state-of-the-art coding agent that relies on a general-purpose shell. GradleFixer's success suggests that while LLMs possess the high-level knowledge to solve these failures, they struggle to translate this knowledge into effective low-level actions using a general-purpose shell. We demonstrate the effectiveness of a strategy we term Tool Bridging, which replaces general-purpose shell commands with domain-aware abstractions. We hypothesize this approach works through two mechanisms: 1) it provides tools in an API-like format that LLMs use more reliably, and 2) it constrains the action space to relevant operations. This approach bridges the gap between the model's high-level reasoning and effective low-level execution.
Authors: Tiancheng Gu, Kaicheng Yang, Kaichen Zhang, Xiang An, Ziyong Feng, Yueyi Zhang, Weidong Cai, Jiankang Deng, Lidong Bing
Abstract: Universal multimodal embedding models are foundational to various tasks. Existing approaches typically employ in-batch negative mining by measuring the similarity of query-candidate pairs. However, these methods often struggle to capture subtle semantic differences among candidates and lack diversity in negative samples. Moreover, the embeddings exhibit limited discriminative ability in distinguishing false and hard negatives. In this paper, we leverage the advanced understanding capabilities of MLLMs to enhance representation learning and present a novel Universal Multimodal Embedding (UniME-V2) model. Our approach first constructs a potential hard negative set through global retrieval. We then introduce the MLLM-as-a-Judge mechanism, which utilizes MLLMs to assess the semantic alignment of query-candidate pairs and generate soft semantic matching scores. These scores serve as a foundation for hard negative mining, mitigating the impact of false negatives and enabling the identification of diverse, high-quality hard negatives. Furthermore, the semantic matching scores are used as soft labels to mitigate the rigid one-to-one mapping constraint. By aligning the similarity matrix with the soft semantic matching score matrix, the model learns semantic distinctions among candidates, significantly enhancing its discriminative capacity. To further improve performance, we propose UniME-V2-Reranker, a reranking model trained on our mined hard negatives through a joint pairwise and listwise optimization approach. We conduct comprehensive experiments on the MMEB benchmark and multiple retrieval tasks, demonstrating that our method achieves state-of-the-art performance on average across all tasks.
Authors: Kun Lei, Huanyu Li, Dongjie Yu, Zhenyu Wei, Lingxiao Guo, Zhennan Jiang, Ziyu Wang, Shiyu Liang, Huazhe Xu
Abstract: Real-world robotic manipulation in homes and factories demands reliability, efficiency, and robustness that approach or surpass the performance of skilled human operators. We present RL-100, a real-world reinforcement learning framework built on diffusion-based visuomotor policies. RL-100 unifies imitation and reinforcement learning under a single PPO-style objective applied within the denoising process, yielding conservative and stable policy improvements across both offline and online stages. To meet deployment latency constraints, we employ a lightweight consistency distillation procedure that compresses multi-step diffusion into a one-step controller for high-frequency control. The framework is task-, embodiment-, and representation-agnostic, and supports both single-action outputs and action-chunking control. We evaluate RL-100 on seven diverse real-robot manipulation tasks, ranging from dynamic pushing and agile bowling to pouring, cloth folding, unscrewing, and multi-stage juicing. RL-100 attains 100% success across evaluated trials, achieving 900 out of 900 successful episodes, including up to 250 out of 250 consecutive trials on one task, and matches or surpasses expert teleoperators in time-to-completion. Without retraining, a single policy attains approximately 90% zero-shot success under environmental and dynamics shifts, adapts in a few-shot regime to significant task variations (86.7%), and remains robust to aggressive human perturbations (about 95%). In a public shopping-mall deployment, the juicing robot served random customers continuously for roughly seven hours without failure. Together, these results suggest a practical path toward deployment-ready robot learning: start from human priors, align training objectives with human-grounded metrics, and reliably extend performance beyond human demonstrations.
Authors: Austin Xu, Xuan-Phi Nguyen, Yilun Zhou, Chien-Sheng Wu, Caiming Xiong, Shafiq Joty
Abstract: Finetuning specialized generative evaluators has emerged as a popular paradigm to meet the increasing demand for scalable evaluation during both training and test-time. However, recent work has largely focused on applying new methodology, such as reinforcement learning (RL), to training evaluators, shying away from large-scale, data-driven development. In this work, we focus on data scaling, curating a set of 2.5M samples spanning five unique evaluation tasks (pairwise, step-level, reference-free and reference-based verification, and single rating) and multiple domains focused on reasoning evaluation. With our data, we train Foundational Automatic Reasoning Evaluators (FARE), a family of 8B and 20B (with 3.6B active) parameter evaluators, with a simple iterative rejection-sampling supervised finetuning (SFT) approach. FARE-8B challenges larger specialized RL-trained evaluators and FARE-20B sets the new standard for open-source evaluators, surpassing specialized 70B+ evaluators. Beyond static benchmarks, we evaluate FARE in real-world tasks: As inference-time rerankers, FARE-20B achieves near-oracle performance on MATH. As verifiers in RL training, FARE improves the downstream RL-trained model performance by up to 14.1% vs. string-matching verifiers. When initialized from FARE, a continually-finetuned FARE-Code outperforms gpt-oss-20B by 65% on evaluating test-case quality.
Authors: Yajie Li, Albert Galimov, Mitra Datta Ganapaneni, Pujitha Thejaswi, De Meng, Priyanshu Kumar, Saloni Potdar
Abstract: Entity Linking (EL) has traditionally relied on large annotated datasets and extensive model fine-tuning. While recent few-shot methods leverage large language models (LLMs) through prompting to reduce training requirements, they often suffer from inefficiencies due to expensive LLM-based reasoning. ARTER (Adaptive Routing and Targeted Entity Reasoning) presents a structured pipeline that achieves high performance without deep fine-tuning by strategically combining candidate generation, context-based scoring, adaptive routing, and selective reasoning. ARTER computes a small set of complementary signals(both embedding and LLM-based) over the retrieved candidates to categorize contextual mentions into easy and hard cases. The cases are then handled by a low-computational entity linker (e.g. ReFinED) and more expensive targeted LLM-based reasoning respectively. On standard benchmarks, ARTER outperforms ReFinED by up to +4.47%, with an average gain of +2.53% on 5 out of 6 datasets, and performs comparably to pipelines using LLM-based reasoning for all mentions, while being as twice as efficient in terms of the number of LLM tokens.
Authors: Jinchang Luo, Mingquan Cheng, Fan Wan, Ni Li, Xiaoling Xia, Shuangshuang Tian, Tingcheng Bian, Haiwei Wang, Haohuan Fu, Yan Tao
Abstract: Reinforcement learning has recently shown promise in improving retrieval-augmented generation (RAG). Despite these advances, its effectiveness in multi-hop question answering (QA) remains limited by two fundamental limitations: (i) global planning absence to structure multi-step reasoning, and (ii) unfaithful execution, which hinders effective query formulation and consistent use of retrieved evidence. We propose GlobalRAG, a reinforcement learning framework designed to enhance global reasoning in multi-hop QA. GlobalRAG decomposes questions into subgoals, coordinates retrieval with reasoning, and refines evidence iteratively. To guide this process, we introduce Planning Quality Reward and SubGoal Completion Reward, which encourage coherent planning and reliable subgoal execution. In addition, a progressive weight annealing strategy balances process-oriented and outcome-based objectives. Extensive experiments on both in-domain and out-of-domain benchmarks demonstrate that GlobalRAG significantly outperforms strong baselines while using only 8k training data (42% of the training data used by strong baselines), achieving average improvements of 14.2% in both EM and F1.
Authors: Kun Chen, Peng Shi, Haibo Qiu, Zhixiong Zeng, Siqi Yang, Wenji Mao, Lin Ma
Abstract: Reinforcement learning (RL) with verifiable rewards has recently catalyzed a wave of "MLLM-r1" approaches that bring RL to vision language models. Most representative paradigms begin with a cold start, typically employing supervised fine-tuning (SFT), to initialize the policy before RL. However, SFT-based cold start adopts the reasoning paradigm intertwined with task solution and output format, which may induce instruction-style overfitting, weakens out-of-distribution generalization, and ultimately affects downstream RL. We revisit the cold start along two views, its training method and data construction, and introduce the Generalization Factor (GF) coefficient to quantify the generalization capability under different methods. Our empirical study finds that preference-based training methods (e.g. DPO) generalizes better than SFT-based methods in cold start. Motivated by this, we propose SPECS-a Self-distilled, Preference-based Cold Start framework that decouples multimodal learning: (1) generates introspective preference data pairs via self-distillation, avoiding reliance on larger teachers or manual annotation; (2) performs preference-based training to learn, focusing on shallow, transferable surface-form criteria (format, structure, style) rather than memorizing content; and (3) hands off to RL with verifiable rewards for deep reasoning results. Experimental results across multiple multimodal benchmarks show that our decoupling learning framework yields consistent performance gains over strong baselines, improving MEGA-Bench by 4.1% and MathVista by 12.2%. Additional experiments indicate that SPECS contributes to reducing in-distribution "stuckness," improving exploration, stabilizing training, and raising the performance ceiling.
Authors: Junyao Shi, Rujia Yang, Kaitian Chao, Selina Bingqing Wan, Yifei Shao, Jiahui Lei, Jianing Qian, Long Le, Pratik Chaudhari, Kostas Daniilidis, Chuan Wen, Dinesh Jayaraman
Abstract: Today's best-explored routes towards generalist robots center on collecting ever larger "observations-in actions-out" robotics datasets to train large end-to-end models, copying a recipe that has worked for vision-language models (VLMs). We pursue a road less traveled: building generalist policies directly around VLMs by augmenting their general capabilities with specific robot capabilities encapsulated in a carefully curated set of perception, planning, and control modules. In Maestro, a VLM coding agent dynamically composes these modules into a programmatic policy for the current task and scenario. Maestro's architecture benefits from a streamlined closed-loop interface without many manually imposed structural constraints, and a comprehensive and diverse tool repertoire. As a result, it largely surpasses today's VLA models for zero-shot performance on challenging manipulation skills. Further, Maestro is easily extensible to incorporate new modules, easily editable to suit new embodiments such as a quadruped-mounted arm, and even easily adapts from minimal real-world experiences through local code edits.
Authors: Jiahui Gao, Kuang Zhou, Yuchen Zhu, Keyu Wu
Abstract: Understanding and quantifying node importance is a fundamental problem in network science and engineering, underpinning a wide range of applications such as influence maximization, social recommendation, and network dismantling. Prior research often relies on centrality measures or advanced graph embedding techniques using structural information, followed by downstream classification or regression tasks to identify critical nodes. However, these methods typically decouple node representation learning from the ranking objective and rely on the topological structure of target networks, leading to feature-task inconsistency and limited generalization across networks. This paper proposes a novel framework that leverages causal representation learning to get robust, invariant node embeddings for cross-network ranking tasks. Firstly, we introduce an influence-aware causal node embedding module within an autoencoder architecture to extract node embeddings that are causally related to node importance. Moreover, we introduce a causal ranking loss and design a unified optimization framework that jointly optimizes the reconstruction and ranking objectives, enabling mutual reinforcement between node representation learning and ranking optimization. This design allows the proposed model to be trained on synthetic networks and to generalize effectively across diverse real-world networks. Extensive experiments on multiple benchmark datasets demonstrate that the proposed model consistently outperforms state-of-the-art baselines in terms of both ranking accuracy and cross-network transferability, offering new insights for network analysis and engineering applications-particularly in scenarios where the target network's structure is inaccessible in advance due to privacy or security constraints.
Authors: Yuxiao Yang, Xiao-Xiao Long, Zhiyang Dou, Cheng Lin, Yuan Liu, Qingsong Yan, Yuexin Ma, Haoqian Wang, Zhiqiang Wu, Wei Yin
Abstract: In this work, we introduce \textbf{Wonder3D++}, a novel method for efficiently generating high-fidelity textured meshes from single-view images. Recent methods based on Score Distillation Sampling (SDS) have shown the potential to recover 3D geometry from 2D diffusion priors, but they typically suffer from time-consuming per-shape optimization and inconsistent geometry. In contrast, certain works directly produce 3D information via fast network inferences, but their results are often of low quality and lack geometric details. To holistically improve the quality, consistency, and efficiency of single-view reconstruction tasks, we propose a cross-domain diffusion model that generates multi-view normal maps and the corresponding color images. To ensure the consistency of generation, we employ a multi-view cross-domain attention mechanism that facilitates information exchange across views and modalities. Lastly, we introduce a cascaded 3D mesh extraction algorithm that drives high-quality surfaces from the multi-view 2D representations in only about $3$ minute in a coarse-to-fine manner. Our extensive evaluations demonstrate that our method achieves high-quality reconstruction results, robust generalization, and good efficiency compared to prior works. Code available at https://github.com/xxlong0/Wonder3D/tree/Wonder3D_Plus.
URLs: https://github.com/xxlong0/Wonder3D/tree/Wonder3D_Plus.
Authors: Chao Yan (Tony), Boyong Wu (Tony), Peng Yang (Tony), Pengfei Tan (Tony), Guoqiang Hu (Tony), Li Xie (Tony), Yuxin Zhang (Tony), Xiangyu (Tony), Zhang, Fei Tian, Xuerui Yang, Xiangyu Zhang, Daxin Jiang, Shuchang Zhou, Gang Yu
Abstract: We present Step-Audio-EditX, the first open-source LLM-based audio model excelling at expressive and iterative audio editing encompassing emotion, speaking style, and paralinguistics alongside robust zero-shot text-to-speech (TTS) capabilities. Our core innovation lies in leveraging only large-margin synthetic data, which circumvents the need for embedding-based priors or auxiliary modules. This large-margin learning approach enables both iterative control and high expressivity across voices, and represents a fundamental pivot from the conventional focus on representation-level disentanglement. Evaluation results demonstrate that Step-Audio-EditX surpasses both MiniMax-2.6-hd and Doubao-Seed-TTS-2.0 in emotion editing and other fine-grained control tasks.
Authors: Davide Marincione, Donato Crisostomi, Roberto Dessi, Emanuele Rodol\`a, Emanuele Rossi
Abstract: Foundation models capable of generalizing across species and tasks represent a promising new frontier in bioacoustics, with NatureLM being one of the most prominent examples. While its domain-specific fine-tuning yields strong performance on bioacoustic benchmarks, we observe that it also introduces trade-offs in instruction-following flexibility. For instance, NatureLM achieves high accuracy when prompted for either the common or scientific name individually, but its accuracy drops significantly when both are requested in a single prompt. We address this by applying a simple model merging strategy that interpolates NatureLM with its base language model, recovering instruction-following capabilities with minimal loss of domain expertise. Finally, we show that the merged model exhibits markedly stronger zero-shot generalization, achieving over a 200% relative improvement and setting a new state-of-the-art in closed-set zero-shot classification of unseen species.
Authors: Richard Goldman, Varun Komperla, Thomas Ploetz, Harish Haresamudram
Abstract: A promising alternative to the computationally expensive Neural Architecture Search (NAS) involves the development of Zero Cost Proxies (ZCPs), which correlate well with trained performance, but can be computed through a single forward/backward pass on a randomly sampled batch of data. In this paper, we investigate the effectiveness of ZCPs for HAR on six benchmark datasets, and demonstrate that they discover network architectures that obtain within 5% of performance attained by full-scale training involving 1500 randomly sampled architectures. This results in substantial computational savings as high-performing architectures can be discovered with minimal training. Our experiments not only introduce ZCPs to sensor-based HAR, but also demonstrate that they are robust to data noise, further showcasing their suitability for practical scenarios.
Authors: Jiangwen Dong, Zehui Lin, Wanyu Lin, Mingjin Zhang
Abstract: Large Language Models (LLMs) have achieved impressive performance in complex reasoning problems. Their effectiveness highly depends on the specific nature of the task, especially the required domain knowledge. Existing approaches, such as mixture-of-experts, typically operate at the task level; they are too coarse to effectively solve the heterogeneous problems involving multiple subjects. This work proposes a novel framework that performs fine-grained analysis at subject level equipped with a designated multi-agent collaboration strategy for addressing heterogeneous problem reasoning. Specifically, given an input query, we first employ a Graph Neural Network to identify the relevant subjects and infer their interdependencies to generate an \textit{Subject-based Directed Acyclic Graph} (S-DAG), where nodes represent subjects and edges encode information flow. Then we profile the LLM models by assigning each model a subject-specific expertise score, and select the top-performing one for matching corresponding subject of the S-DAG. Such subject-model matching enables graph-structured multi-agent collaboration where information flows from the starting model to the ending model over S-DAG. We curate and release multi-subject subsets of standard benchmarks (MMLU-Pro, GPQA, MedMCQA) to better reflect complex, real-world reasoning tasks. Extensive experiments show that our approach significantly outperforms existing task-level model selection and multi-agent collaboration baselines in accuracy and efficiency. These results highlight the effectiveness of subject-aware reasoning and structured collaboration in addressing complex and multi-subject problems.
Authors: Pawe{\l} Liskowski, Benjamin Han, Paritosh Aggarwal, Bowei Chen, Boxin Jiang, Nitish Jindal, Zihan Li, Aaron Lin, Kyle Schmaus, Jay Tayade, Weicheng Zhao, Anupam Datta, Nathan Wiegand, Dimitris Tsirogiannis
Abstract: Snowflake's Cortex AISQL is a production SQL engine that integrates native semantic operations directly into SQL. This integration allows users to write declarative queries that combine relational operations with semantic reasoning, enabling them to query both structured and unstructured data effortlessly. However, making semantic operations efficient at production scale poses fundamental challenges. Semantic operations are more expensive than traditional SQL operations, possess distinct latency and throughput characteristics, and their cost and selectivity are unknown during query compilation. Furthermore, existing query engines are not designed to optimize semantic operations. The AISQL query execution engine addresses these challenges through three novel techniques informed by production deployment data from Snowflake customers. First, AI-aware query optimization treats AI inference cost as a first-class optimization objective, reasoning about large language model (LLM) cost directly during query planning to achieve 2-8$\times$ speedups. Second, adaptive model cascades reduce inference costs by routing most rows through a fast proxy model while escalating uncertain cases to a powerful oracle model, achieving 2-6$\times$ speedups while maintaining 90-95% of oracle model quality. Third, semantic join query rewriting lowers the quadratic time complexity of join operations to linear through reformulation as multi-label classification tasks, achieving 15-70$\times$ speedups with often improved prediction quality. AISQL is deployed in production at Snowflake, where it powers diverse customer workloads across analytics, search, and content understanding.
Authors: Huijie Liu, Shuhao Cui, Haoxiang Cao, Shuai Ma, Kai Wu, Guoliang Kang
Abstract: Innovative visual stylization is a cornerstone of artistic creation, yet generating novel and consistent visual styles remains a significant challenge. Existing generative approaches typically rely on lengthy textual prompts, reference images, or parameter-efficient fine-tuning to guide style-aware image generation, but often struggle with style consistency, limited creativity, and complex style representations. In this paper, we affirm that a style is worth one numerical code by introducing the novel task, code-to-style image generation, which produces images with novel, consistent visual styles conditioned solely on a numerical style code. To date, this field has only been primarily explored by the industry (e.g., Midjourney), with no open-source research from the academic community. To fill this gap, we propose CoTyle, the first open-source method for this task. Specifically, we first train a discrete style codebook from a collection of images to extract style embeddings. These embeddings serve as conditions for a text-to-image diffusion model (T2I-DM) to generate stylistic images. Subsequently, we train an autoregressive style generator on the discrete style embeddings to model their distribution, allowing the synthesis of novel style embeddings. During inference, a numerical style code is mapped to a unique style embedding by the style generator, and this embedding guides the T2I-DM to generate images in the corresponding style. Unlike existing methods, our method offers unparalleled simplicity and diversity, unlocking a vast space of reproducible styles from minimal input. Extensive experiments validate that CoTyle effectively turns a numerical code into a style controller, demonstrating a style is worth one code.
Authors: Jonathan Teo (Singapore Management University), Lee Xin Wei (Singapore Management University), Hoong Chuin Lau (Singapore Management University)
Abstract: The Equivariant Quantum Circuit (EQC) for the Travelling Salesman Problem (TSP) has been shown to achieve near-optimal performance in solving small TSP problems (up to 20 nodes) using only two parameters at depth 1. However, extending EQCs to larger TSP problem sizes remains challenging due to the exponential time and memory for quantum circuit simulation, as well as increasing noise and decoherence when running on actual quantum hardware. In this work, we propose the Size-Invariant Grid Search (SIGS), an efficient training optimization for Quantum Reinforcement Learning (QRL), and use it to simulate the outputs of a trained Depth-1 EQC up to 350-node TSP instances - well beyond previously tractable limits. At TSP with 100 nodes, we reduce total simulation times by 96.4%, when comparing to RL simulations with the analytical expression (151 minutes using RL to under 6 minutes using SIGS on TSP-100), while achieving a mean optimality gap within 0.005 of the RL trained model on the test set. SIGS provides a practical benchmarking tool for the QRL community, allowing us to efficiently analyze the performance of QRL algorithms on larger problem sizes. We provide a theoretical explanation for SIGS called the Size-Invariant Properties that goes beyond the concept of equivariance discussed in prior literature.
Authors: Luciano Araujo Dourado Filho, Almir Moreira da Silva Neto, Anthony Miyaguchi, Rodrigo Pereira David, Rodrigo Tripodi Calumby, Luk\'a\v{s} Picek
Abstract: This paper proposes a competitive and computationally efficient approach to probabilistic rainfall nowcasting. A video projector (V-JEPA Vision Transformer) associated to a lightweight probabilistic head is attached to a pre-trained satellite vision encoder (DINOv3-SAT493M) to map encoder tokens into a discrete empirical CDF (eCDF) over 4-hour accumulated rainfall. The projector-head is optimized end-to-end over the Ranked Probability Score (RPS). As an alternative, 3D-UNET baselines trained with an aggregate Rank Probability Score and a per-pixel Gamma-Hurdle objective are used. On the Weather4Cast 2025 benchmark, the proposed method achieved a promising performance, with a CRPS of 3.5102, which represents $\approx$ 26% in effectiveness gain against the best 3D-UNET.
Authors: Xueren Ge, Sahil Murtaza, Anthony Cortez, Homa Alemzadeh
Abstract: Large language models (LLMs) have shown promise in medical question answering, yet they often overlook the domain-specific expertise that professionals depend on, such as the clinical subject areas (e.g., trauma, airway) and the certification level (e.g., EMT, Paramedic). Existing approaches typically apply general-purpose prompting or retrieval strategies without leveraging this structured context, limiting performance in high-stakes settings. We address this gap with EMSQA, an 24.3K-question multiple-choice dataset spanning 10 clinical subject areas and 4 certification levels, accompanied by curated, subject area-aligned knowledge bases (40K documents and 2M tokens). Building on EMSQA, we introduce (i) Expert-CoT, a prompting strategy that conditions chain-of-thought (CoT) reasoning on specific clinical subject area and certification level, and (ii) ExpertRAG, a retrieval-augmented generation pipeline that grounds responses in subject area-aligned documents and real-world patient data. Experiments on 4 LLMs show that Expert-CoT improves up to 2.05% over vanilla CoT prompting. Additionally, combining Expert-CoT with ExpertRAG yields up to a 4.59% accuracy gain over standard RAG baselines. Notably, the 32B expertise-augmented LLMs pass all the computer-adaptive EMS certification simulation exams.
Authors: Yiming Zhang, Yuejia Xu, Ziyao Wang, Xin Yan, Xiaosai Hao
Abstract: Globally, disparities in healthcare infrastructure remain stark, leaving countless communities without access to even basic services. Traditional infrastructure planning is often slow and inaccessible, and although many architects are actively delivering humanitarian and aid-driven hospital projects worldwide, these vital efforts still fall far short of the sheer scale and urgency of demand. This paper introduces MedBuild AI, a hybrid-intelligence framework that integrates large language models (LLMs) with deterministic expert systems to rebalance the early design and conceptual planning stages. As a web-based platform, it enables any region with satellite internet access to obtain guidance on modular, low-tech, low-cost medical building designs. The system operates through three agents: the first gathers local health intelligence via conversational interaction; the second translates this input into an architectural functional program through rule-based computation; and the third generates layouts and 3D models. By embedding computational negotiation into the design process, MedBuild AI fosters a reciprocal, inclusive, and equitable approach to healthcare planning, empowering communities and redefining agency in global healthcare architecture.
Authors: Sebasti\'an Andr\'es Cajas Ord\'o\~nez, Luis Fernando Torres Torres, Mackenzie J. Meni, Carlos Andr\'es Duran Paredes, Eric Arazo, Cristian Bosch, Ricardo Simon Carbajo, Yuan Lai, Leo Anthony Celi
Abstract: Deploying deep neural networks on resource-constrained devices faces two critical challenges: maintaining accuracy under aggressive quantization while ensuring predictable inference latency. We present a curiosity-driven quantized Mixture-of-Experts framework that addresses both through Bayesian epistemic uncertainty-based routing across heterogeneous experts (BitNet ternary, 1-16 bit BitLinear, post-training quantization). Evaluated on audio classification benchmarks (ESC-50, Quinn, UrbanSound8K), our 4-bit quantization maintains 99.9 percent of 16-bit accuracy (0.858 vs 0.859 F1) with 4x compression and 41 percent energy savings versus 8-bit. Crucially, curiosity-driven routing reduces MoE latency variance by 82 percent (p = 0.008, Levene's test) from 230 ms to 29 ms standard deviation, enabling stable inference for battery-constrained devices. Statistical analysis confirms 4-bit/8-bit achieve practical equivalence with full precision (p > 0.05), while MoE architectures introduce 11 percent latency overhead (p < 0.001) without accuracy gains. At scale, deployment emissions dominate training by 10000x for models serving more than 1,000 inferences, making inference efficiency critical. Our information-theoretic routing demonstrates that adaptive quantization yields accurate (0.858 F1, 1.2M params), energy-efficient (3.87 F1/mJ), and predictable edge models, with simple 4-bit quantized architectures outperforming complex MoE for most deployments.
Authors: Zhengxin Zhang, Chengyu Huang, Aochong Oliver Li, Claire Cardie
Abstract: Large Language Models (LLMs) have achieved remarkable progress through Reinforcement Learning with Verifiable Rewards (RLVR), yet still rely heavily on external supervision (e.g., curated labels). Adversarial learning, particularly through self-play, offers a promising alternative that enables models to iteratively learn from themselves - thus reducing reliance on external supervision. Dual-play extends adversarial learning by assigning specialized roles to two models and training them against each other, fostering sustained competition and mutual evolution. Despite its promise, adapting dual-play training to LLMs remains limited, largely due to their susceptibility to reward hacking and training instability. In this paper, we introduce PasoDoble, a novel LLM dual-play framework. PasoDoble adversarially trains two models initialized from the same base model: a Proposer, which generates challenging questions with ground-truth answers, and a Solver, which attempts to solve them. We enrich the Proposer with knowledge from a pre-training dataset to ensure the questions' quality and diversity. To avoid reward hacking, the Proposer is rewarded for producing only valid questions that push the Solver's limit, while the Solver is rewarded for solving them correctly, and both are updated jointly. To further enhance training stability, we introduce an optional offline paradigm that decouples Proposer and Solver updates, alternately updating each for several steps while holding the other fixed. Notably, PasoDoble operates without supervision during training. Experimental results show that PasoDoble can improve the reasoning performance of LLMs. Our project page is available at https://hcy123902.github.io/PasoDoble.
Authors: Xinyue Yu, Youqing Fang, Pingyu Wu, Guoyang Ye, Wenbo Zhou, Weiming Zhang, Song Xiao
Abstract: Generating expressive and controllable human speech is one of the core goals of generative artificial intelligence, but its progress has long been constrained by two fundamental challenges: the deep entanglement of speech factors and the coarse granularity of existing control mechanisms. To overcome these challenges, we have proposed a novel framework called MF-Speech, which consists of two core components: MF-SpeechEncoder and MF-SpeechGenerator. MF-SpeechEncoder acts as a factor purifier, adopting a multi-objective optimization strategy to decompose the original speech signal into highly pure and independent representations of content, timbre, and emotion. Subsequently, MF-SpeechGenerator functions as a conductor, achieving precise, composable and fine-grained control over these factors through dynamic fusion and Hierarchical Style Adaptive Normalization (HSAN). Experiments demonstrate that in the highly challenging multi-factor compositional speech generation task, MF-Speech significantly outperforms current state-of-the-art methods, achieving a lower word error rate (WER=4.67%), superior style control (SECS=0.5685, Corr=0.68), and the highest subjective evaluation scores(nMOS=3.96, sMOS_emotion=3.86, sMOS_style=3.78). Furthermore, the learned discrete factors exhibit strong transferability, demonstrating their significant potential as a general-purpose speech representation.
Authors: Taiyi Su, Jian Zhu, Yaxuan Li, Chong Ma, Zitai Huang, Hanli Wang, Yi Xu
Abstract: Embodied world models aim to predict and interact with the physical world through visual observations and actions. However, existing models struggle to accurately translate low-level actions (e.g., joint positions) into precise robotic movements in predicted frames, leading to inconsistencies with real-world physical interactions. To address these limitations, we propose MTV-World, an embodied world model that introduces Multi-view Trajectory-Video control for precise visuomotor prediction. Specifically, instead of directly using low-level actions for control, we employ trajectory videos obtained through camera intrinsic and extrinsic parameters and Cartesian-space transformation as control signals. However, projecting 3D raw actions onto 2D images inevitably causes a loss of spatial information, making a single view insufficient for accurate interaction modeling. To overcome this, we introduce a multi-view framework that compensates for spatial information loss and ensures high-consistency with physical world. MTV-World forecasts future frames based on multi-view trajectory videos as input and conditioning on an initial frame per view. Furthermore, to systematically evaluate both robotic motion precision and object interaction accuracy, we develop an auto-evaluation pipeline leveraging multimodal large models and referring video object segmentation models. To measure spatial consistency, we formulate it as an object location matching problem and adopt the Jaccard Index as the evaluation metric. Extensive experiments demonstrate that MTV-World achieves precise control execution and accurate physical interaction modeling in complex dual-arm scenarios.
Authors: Rasmus F. Orsoe, Stephan Meighen-Berger, Jeffrey Lazar, Jorge Prado, Ivan Mozun-Mateo, Aske Rosted, Philip Weigel, Arturo Llorente Anaya
Abstract: Neutrino telescopes are large-scale detectors designed to observe Cherenkov radiation produced from neutrino interactions in water or ice. They exist to identify extraterrestrial neutrino sources and to probe fundamental questions pertaining to the elusive neutrino itself. A central challenge common across neutrino telescopes is to solve a series of inverse problems known as event reconstruction, which seeks to resolve properties of the incident neutrino, based on the detected Cherenkov light. In recent times, significant efforts have been made in adapting advances from deep learning research to event reconstruction, as such techniques provide several benefits over traditional methods. While a large degree of similarity in reconstruction needs and low-level data exists, cross-experimental collaboration has been hindered by a lack of diverse open-source datasets for comparing methods. We present NuBench, an open benchmark for deep learning-based event reconstruction in neutrino telescopes. NuBench comprises seven large-scale simulated datasets containing nearly 130 million charged- and neutral-current muon-neutrino interactions spanning 10 GeV to 100 TeV, generated across six detector geometries inspired by existing and proposed experiments. These datasets provide pulse- and event-level information suitable for developing and comparing machine-learning reconstruction methods in both water and ice environments. Using NuBench, we evaluate four reconstruction algorithms - ParticleNeT and DynEdge, both actively used within the KM3NeT and IceCube collaborations, respectively, along with GRIT and DeepIce - on up to five core tasks: energy and direction reconstruction, topology classification, interaction vertex prediction, and inelasticity estimation.
Authors: Amirreza Mehrabi, Jason Wade Morphew, Breejha Quezada, N. Sanjay Rebello
Abstract: Adaptive learning often diagnoses precisely yet intervenes weakly, producing help that is mistimed or misaligned. This study presents evidence supporting an instructor-governed feedback loop that converts concept-level assessment evidence into vetted microinterventions. The adaptive learning algorithm includes three safeguards: adequacy as a hard guarantee of gap closure, attention as a budgeted limit for time and redundancy, and diversity as protection against overfitting to a single resource. We formulate intervention assignment as a binary integer program with constraints for coverage, time, difficulty windows derived from ability estimates, prerequisites encoded by a concept matrix, and anti-redundancy with diversity. Greedy selection serves low-richness and tight-latency settings, gradient-based relaxation serves rich repositories, and a hybrid switches along a richness-latency frontier. In simulation and in an introductory physics deployment with 1204 students, both solvers achieved full skill coverage for nearly all learners within bounded watch time. The gradient-based method reduced redundant coverage by about 12 percentage points relative to greedy and produced more consistent difficulty alignment, while greedy delivered comparable adequacy at lower computational cost in resource-scarce environments. Slack variables localized missing content and guided targeted curation, sustaining sufficiency across student subgroups. The result is a tractable and auditable controller that closes the diagnostic pedagogical loop and enables equitable, load-aware personalization at the classroom scale.
Authors: Dimitrios Koutsianos, Ladislav Mosner, Yannis Panagakis, Themos Stafylakis
Abstract: Performance in face and speaker verification is largely driven by margin based softmax losses like CosFace and ArcFace. Recently introduced $\alpha$-divergence loss functions offer a compelling alternative, particularly for their ability to induce sparse solutions (when $\alpha>1$). However, integrating an angular margin-crucial for verification tasks-is not straightforward. We find this integration can be achieved in at least two distinct ways: via the reference measure (prior probabilities) or via the logits (unnormalized log-likelihoods). In this paper, we explore both pathways, deriving two novel margin-based $\alpha$-divergence losses: Q-Margin (margin in the reference measure) and A3M (margin in the logits). We identify and address a critical training instability in A3M-caused by the interplay of penalized logits and sparsity-with a simple yet effective prototype re-initialization strategy. Our methods achieve significant performance gains on the challenging IJB-B and IJB-C face verification benchmarks. We demonstrate similarly strong performance in speaker verification on VoxCeleb. Crucially, our models significantly outperform strong baselines at low false acceptance rates (FAR). This capability is crucial for practical high-security applications, such as banking authentication, when minimizing false authentications is paramount.
Authors: Xueyang Li, Zongren Wang, Yuliang Zhang, Zixuan Pan, Yu-Jen Chen, Nishchal Sapkota, Gelei Xu, Danny Z. Chen, Yiyu Shi
Abstract: Bladder cancer is one of the most prevalent malignancies worldwide, with a recurrence rate of up to 78%, necessitating accurate post-operative monitoring for effective patient management. Multi-sequence contrast-enhanced MRI is commonly used for recurrence detection; however, interpreting these scans remains challenging, even for experienced radiologists, due to post-surgical alterations such as scarring, swelling, and tissue remodeling. AI-assisted diagnostic tools have shown promise in improving bladder cancer recurrence prediction, yet progress in this field is hindered by the lack of dedicated multi-sequence MRI datasets for recurrence assessment study. In this work, we first introduce a curated multi-sequence, multi-modal MRI dataset specifically designed for bladder cancer recurrence prediction, establishing a valuable benchmark for future research. We then propose H-CNN-ViT, a new Hierarchical Gated Attention Multi-Branch model that enables selective weighting of features from the global (ViT) and local (CNN) paths based on contextual demands, achieving a balanced and targeted feature fusion. Our multi-branch architecture processes each modality independently, ensuring that the unique properties of each imaging channel are optimally captured and integrated. Evaluated on our dataset, H-CNN-ViT achieves an AUC of 78.6%, surpassing state-of-the-art models. Our model is publicly available at https://github.com/XLIAaron/H-CNN-ViT.
Authors: Chenchen Kuai, Zihao Li, Braden Rosen, Stephanie Paal, Navid Jafari, Jean-Louis Briaud, Yunlong Zhang, Youssef M. A. Hashash, Yang Zhou
Abstract: Post-disaster reconnaissance reports contain critical evidence for understanding multi-hazard interactions, yet their unstructured narratives make systematic knowledge transfer difficult. Large language models (LLMs) offer new potential for analyzing these reports, but often generate unreliable or hallucinated outputs when domain grounding is absent. This study introduces the Mixture-of-Retrieval Agentic RAG (MoRA-RAG), a knowledge-grounded LLM framework that transforms reconnaissance reports into a structured foundation for multi-hazard reasoning. The framework integrates a Mixture-of-Retrieval mechanism that dynamically routes queries across hazard-specific databases while using agentic chunking to preserve contextual coherence during retrieval. It also includes a verification loop that assesses evidence sufficiency, refines queries, and initiates targeted searches when information remains incomplete. We construct HazardRecQA by deriving question-answer pairs from GEER reconnaissance reports, which document 90 global events across seven major hazard types. MoRA-RAG achieves up to 94.5 percent accuracy, outperforming zero-shot LLMs by 30 percent and state-of-the-art RAG systems by 10 percent, while reducing hallucinations across diverse LLM architectures. MoRA-RAG also enables open-weight LLMs to achieve performance comparable to proprietary models. It establishes a new paradigm for transforming post-disaster documentation into actionable, trustworthy intelligence for hazard resilience.
Authors: Kangqiao Zhao, Shuo Huai, Xurui Song, Jun Luo
Abstract: Though deep neural models adopted to realize the perception of autonomous driving have proven vulnerable to adversarial examples, known attacks often leverage 2D patches and target mostly monocular perception. Therefore, the effectiveness of Physical Adversarial Examples (PAEs) on stereo-based binocular depth estimation remains largely unexplored. To this end, we propose the first texture-enabled physical adversarial attack against stereo matching models in the context of autonomous driving. Our method employs a 3D PAE with global camouflage texture rather than a local 2D patch-based one, ensuring both visual consistency and attack effectiveness across different viewpoints of stereo cameras. To cope with the disparity effect of these cameras, we also propose a new 3D stereo matching rendering module that allows the PAE to be aligned with real-world positions and headings in binocular vision. We further propose a novel merging attack that seamlessly blends the target into the environment through fine-grained PAE optimization. It has significantly enhanced stealth and lethality upon existing hiding attacks that fail to get seamlessly merged into the background. Extensive evaluations show that our PAEs can successfully fool the stereo models into producing erroneous depth information.
Authors: Soham Ghosh, Gaurav Mittal
Abstract: Agentic AI systems have recently emerged as a critical and transformative approach in artificial intelligence, offering capabilities that extend far beyond traditional AI agents and contemporary generative AI models. This rapid evolution necessitates a clear conceptual and taxonomical understanding to differentiate this new paradigm. Our paper addresses this gap by providing a comprehensive review that establishes a precise definition and taxonomy for "agentic AI," with the aim of distinguishing it from previous AI paradigms. The concepts are gradually introduced, starting with a highlight of its diverse applications across the broader field of engineering. The paper then presents four detailed, state-of-the-art use case applications specifically within electrical engineering. These case studies demonstrate practical impact, ranging from an advanced agentic framework for streamlining complex power system studies and benchmarking to a novel system developed for survival analysis of dynamic pricing strategies in battery swapping stations. Finally, to ensure robust deployment, the paper provides detailed failure mode investigations. From these findings, we derive actionable recommendations for the design and implementation of safe, reliable, and accountable agentic AI systems, offering a critical resource for researchers and practitioners.
Authors: Chen Chen, Cuong Nguyen, Alexa Siu, Dingzeyu Li, Nadir Weibel
Abstract: Accessing 3D models remains challenging for Screen Reader (SR) users. While some existing 3D viewers allow creators to provide alternative text, they often lack sufficient detail about the 3D models. Grounded on a formative study, this paper introduces SweeperBot, a system that enables SR users to leverage visual question answering to explore and compare 3D models. SweeperBot answers SR users' visual questions by combining an optimal view selection technique with the strength of generative- and recognition-based foundation models. An expert review with 10 Blind and Low-Vision (BLV) users with SR experience demonstrated the feasibility of using SweeperBot to assist BLV users in exploring and comparing 3D models. The quality of the descriptions generated by SweeperBot was validated by a second survey study with 30 sighted participants.
Authors: Xianhui Meng, Yuchen Zhang, Zhijian Huang, Zheng Lu, Ziling Ji, Yaoyao Yin, Hongyuan Zhang, Guangfeng Jiang, Yandan Lin, Long Chen, Hangjun Ye, Li Zhang, Jun Liu, Xiaoshuai Hao
Abstract: Vision-Language Models (VLMs) show great promise for autonomous driving, but their suitability for safety-critical scenarios is largely unexplored, raising safety concerns. This issue arises from the lack of comprehensive benchmarks that assess both external environmental risks and in-cabin driving behavior safety simultaneously. To bridge this critical gap, we introduce DSBench, the first comprehensive Driving Safety Benchmark designed to assess a VLM's awareness of various safety risks in a unified manner. DSBench encompasses two major categories: external environmental risks and in-cabin driving behavior safety, divided into 10 key categories and a total of 28 sub-categories. This comprehensive evaluation covers a wide range of scenarios, ensuring a thorough assessment of VLMs' performance in safety-critical contexts. Extensive evaluations across various mainstream open-source and closed-source VLMs reveal significant performance degradation under complex safety-critical situations, highlighting urgent safety concerns. To address this, we constructed a large dataset of 98K instances focused on in-cabin and external safety scenarios, showing that fine-tuning on this dataset significantly enhances the safety performance of existing VLMs and paves the way for advancing autonomous driving technology. The benchmark toolkit, code, and model checkpoints will be publicly accessible.
Authors: Abolfazl Younesi, Leon Kiss, Zahra Najafabadi Samani, Juan Aznar Poveda, Thomas Fahringer
Abstract: Federated learning (FL) enables collaborative model training while preserving data privacy. However, it remains vulnerable to malicious clients who compromise model integrity through Byzantine attacks, data poisoning, or adaptive adversarial behaviors. Existing defense mechanisms rely on static thresholds and binary classification, failing to adapt to evolving client behaviors in real-world deployments. We propose FLARE, an adaptive reputation-based framework that transforms client reliability assessment from binary decisions to a continuous, multi-dimensional trust evaluation. FLARE integrates: (i) a multi-dimensional reputation score capturing performance consistency, statistical anomaly indicators, and temporal behavior, (ii) a self-calibrating adaptive threshold mechanism that adjusts security strictness based on model convergence and recent attack intensity, (iii) reputation-weighted aggregation with soft exclusion to proportionally limit suspicious contributions rather than eliminating clients outright, and (iv) a Local Differential Privacy (LDP) mechanism enabling reputation scoring on privatized client updates. We further introduce a highly evasive Statistical Mimicry (SM) attack, a benchmark adversary that blends honest gradients with synthetic perturbations and persistent drift to remain undetected by traditional filters. Extensive experiments with 100 clients on MNIST, CIFAR-10, and SVHN demonstrate that FLARE maintains high model accuracy and converges faster than state-of-the-art Byzantine-robust methods under diverse attack types, including label flipping, gradient scaling, adaptive attacks, ALIE, and SM. FLARE improves robustness by up to 16% and preserves model convergence within 30% of the non-attacked baseline, while achieving strong malicious-client detection performance with minimal computational overhead. https://github.com/Anonymous0-0paper/FLARE