Authors: Luciano Floridi
Abstract: This article introduces a conjecture that formalises a fundamental trade-off between provable correctness and broad data-mapping capacity in Artificial Intelligence (AI) systems. When an AI system is engineered for deductively watertight guarantees (demonstrable certainty about the error-free nature of its outputs) -- as in classical symbolic AI -- its operational domain must be narrowly circumscribed and pre-structured. Conversely, a system that can input high-dimensional data to produce rich information outputs -- as in contemporary generative models -- necessarily relinquishes the possibility of zero-error performance, incurring an irreducible risk of errors or misclassification. By making this previously implicit trade-off explicit and open to rigorous verification, the conjecture significantly reframes both engineering ambitions and philosophical expectations for AI. After reviewing the historical motivations for this tension, the article states the conjecture in information-theoretic form and contextualises it within broader debates in epistemology, formal verification, and the philosophy of technology. It then offers an analysis of its implications and consequences, drawing on notions of underdetermination, prudent epistemic risk, and moral responsibility. The discussion clarifies how, if correct, the conjecture would help reshape evaluation standards, governance frameworks, and hybrid system design. The conclusion underscores the importance of eventually proving or refuting the inequality for the future of trustworthy AI.
Authors: Michelle M. Li, Ben Y. Reis, Adam Rodman, Tianxi Cai, Noa Dagan, Ran D. Balicer, Joseph Loscalzo, Isaac S. Kohane, Marinka Zitnik
Abstract: Medical foundation models, including language models trained on clinical notes, vision-language models on medical images, and multimodal models on electronic health records, can summarize clinical notes, answer medical questions, and assist in decision-making. Adapting these models to new populations, specialties, or settings typically requires fine-tuning, careful prompting, or retrieval from knowledge bases. This can be impractical, and limits their ability to interpret unfamiliar inputs and adjust to clinical situations not represented during training. As a result, models are prone to contextual errors, where predictions appear reasonable but fail to account for critical patient-specific or contextual information. These errors stem from a fundamental limitation that current models struggle with: dynamically adjusting their behavior across evolving contexts of medical care. In this Perspective, we outline a vision for context-switching in medical AI: models that dynamically adapt their reasoning without retraining to new specialties, populations, workflows, and clinical roles. We envision context-switching AI to diagnose, manage, and treat a wide range of diseases across specialties and regions, and expand access to medical care.
Authors: Hamzah Dabool, Raghad Mustafa
Abstract: Understanding the distinction between causation and correlation is critical in Alzheimer's disease (AD) research, as it impacts diagnosis, treatment, and the identification of true disease drivers. This experiment investigates the relationships among clinical, cognitive, genetic, and biomarker features using a combination of correlation analysis, machine learning classification, and model interpretability techniques. Employing the XGBoost algorithm, we identified key features influencing AD classification, including cognitive scores and genetic risk factors. Correlation matrices revealed clusters of interrelated variables, while SHAP (SHapley Additive exPlanations) values provided detailed insights into feature contributions across disease stages. Our results highlight that strong correlations do not necessarily imply causation, emphasizing the need for careful interpretation of associative data. By integrating feature importance and interpretability with classical statistical analysis, this work lays groundwork for future causal inference studies aimed at uncovering true pathological mechanisms. Ultimately, distinguishing causal factors from correlated markers can lead to improved early diagnosis and targeted interventions for Alzheimer's disease.
Authors: Filip Cano
Abstract: Ensuring responsible use of artificial intelligence (AI) has become imperative as autonomous systems increasingly influence critical societal domains. However, the concept of trustworthy AI remains broad and multi-faceted. This thesis advances knowledge in the safety, fairness, transparency, and accountability of AI systems. In safety, we extend classical deterministic shielding techniques to become resilient against delayed observations, enabling practical deployment in real-world conditions. We also implement both deterministic and probabilistic safety shields into simulated autonomous vehicles to prevent collisions with road users, validating the use of these techniques in realistic driving simulators. We introduce fairness shields, a novel post-processing approach to enforce group fairness in sequential decision-making settings over finite and periodic time horizons. By optimizing intervention costs while strictly ensuring fairness constraints, this method efficiently balances fairness with minimal interference. For transparency and accountability, we propose a formal framework for assessing intentional behaviour in probabilistic decision-making agents, introducing quantitative metrics of agency and intention quotient. We use these metrics to propose a retrospective analysis of intention, useful for determining responsibility when autonomous systems cause unintended harm. Finally, we unify these contributions through the ``reactive decision-making'' framework, providing a general formalization that consolidates previous approaches. Collectively, the advancements presented contribute practically to the realization of safer, fairer, and more accountable AI systems, laying the foundations for future research in trustworthy AI.
Authors: Qiyue Yin, Pei Xu, Qiaozhe Li, Shengda Liu, Shengqi Shen, Tong Wang, Yihong Han, Xiaonan Zhao, Likun Yang, Shiyue Cao, Shiyu Qiu, Yuxuan Liu, Shizhao Yu, Lei Cui, Chengxin Yan, Jie Sun, Xiangquan Tang, Kaiqi Huang
Abstract: Recent breakthroughs in Large Language Models (LLMs) have led to a qualitative leap in artificial intelligence' s performance on reasoning tasks, particularly demonstrating remarkable capabilities in mathematical, symbolic, and commonsense reasoning. However, as a critical component of advanced human cognition, strategic reasoning, i.e., the ability to assess multi-agent behaviors in dynamic environments, formulate action plans, and adapt strategies, has yet to be systematically evaluated or modeled. To address this gap, this paper introduces WGSR-Bench, the first strategy reasoning benchmark for LLMs using wargame as its evaluation environment. Wargame, a quintessential high-complexity strategic scenario, integrates environmental uncertainty, adversarial dynamics, and non-unique strategic choices, making it an effective testbed for assessing LLMs' capabilities in multi-agent decision-making, intent inference, and counterfactual reasoning. WGSR-Bench designs test samples around three core tasks, i.e., Environmental situation awareness, Opponent risk modeling and Policy generation, which serve as the core S-POE architecture, to systematically assess main abilities of strategic reasoning. Finally, an LLM-based wargame agent is designed to integrate these parts for a comprehensive strategy reasoning assessment. With WGSR-Bench, we hope to assess the strengths and limitations of state-of-the-art LLMs in game-theoretic strategic reasoning and to advance research in large model-driven strategic intelligence.
Authors: Xinmin Fang, Lingfeng Tao, Zhengxiong Li
Abstract: Artificial Intelligence (AI) is reframed as a cognitive engine driving a novel productivity revolution distinct from the Industrial Revolution's physical thrust. This paper develops a theoretical framing of AI as a cognitive revolution akin to written language - a transformative augmentation of human intellect rather than another mechanized tool. We compare AI's emergence to historical leaps in information technology to show how it amplifies knowledge work. Examples from various domains demonstrate AI's impact as a driver of productivity in cognitive tasks. We adopt a multidisciplinary perspective combining computer science advances with economic insights and sociological perspectives on how AI reshapes work and society. Through conceptual frameworks, we visualize the shift from manual to cognitive productivity. Our central argument is that AI functions as an engine of cognition - comparable to how human language revolutionized knowledge - heralding a new productivity paradigm. We discuss how this revolution demands rethinking of skills, organizations, and policies. This paper, balancing academic rigor with clarity, concludes that AI's promise lies in complementing human cognitive abilities, marking a new chapter in productivity evolution.
Authors: Jasper Yao
Abstract: We establish fundamental computational complexity barriers to verifying AI safety as system capabilities scale. Our main results show that for AI systems with expressiveness EXP$(m)$ above a critical threshold $\tau$, safety verification requires exponential time and is coNP-complete. We formalize the Capability-Risk Scaling (CRS) dynamic, which demonstrates how increasing AI capability drives societal safety requirements toward perfection, creating an inescapable tension with verification complexity. Through four core theorems, we prove that (1) verification complexity grows exponentially with system expressiveness, (2) safe policies comprise at most a $2^{-2^m}$ fraction of the policy space, (3) no finite set of alignment techniques can provide universal coverage, and (4) robust safety properties form measure-zero sets for neural networks. These results characterize an "intractability gap" where practical safety requirements fall within the region of computational intractability. We conclude by presenting a strategic trilemma: AI development must either constrain system complexity to maintain verifiable safety, accept unverifiable risks while scaling capabilities, or develop fundamentally new safety paradigms beyond verification. Our work provides the first systematic complexity-theoretic analysis of AI alignment and establishes rigorous bounds that any safety approach must confront. A formal verification of the core theorems in Lean4 is currently in progress.
Authors: Cameron Angliss, Jiaxun Cui, Jiaheng Hu, Arrasy Rahman, Peter Stone
Abstract: Developing AI agents that can robustly adapt to dramatically different strategic landscapes without retraining is a central challenge for multi-agent learning. Pok\'emon Video Game Championships (VGC) is a domain with an extraordinarily large space of possible team configurations of approximately $10^{139}$ - far larger than those of Dota or Starcraft. The highly discrete, combinatorial nature of team building in Pok\'emon VGC causes optimal strategies to shift dramatically depending on both the team being piloted and the opponent's team, making generalization uniquely challenging. To advance research on this problem, we introduce VGC-Bench: a benchmark that provides critical infrastructure, standardizes evaluation protocols, and supplies human-play datasets and a range of baselines - from large-language-model agents and behavior cloning to reinforcement learning and empirical game-theoretic methods such as self-play, fictitious play, and double oracle. In the restricted setting where an agent is trained and evaluated on a single-team configuration, our methods are able to win against a professional VGC competitor. We extensively evaluated all baseline methods over progressively larger team sets and find that even the best-performing algorithm in the single-team setting struggles at scaling up as team size grows. Thus, policy generalization across diverse team strategies remains an open challenge for the community. Our code is open sourced at https://github.com/cameronangliss/VGC-Bench.
Authors: Zaijing Li, Yuquan Xie, Rui Shao, Gongwei Chen, Weili Guan, Dongmei Jiang, Liqiang Nie
Abstract: Recently, agents based on multimodal large language models (MLLMs) have achieved remarkable progress across various domains. However, building a generalist agent with capabilities such as perception, planning, action, grounding, and reflection in open-world environments like Minecraft remains challenges: insufficient domain-specific data, interference among heterogeneous tasks, and visual diversity in open-world settings. In this paper, we address these challenges through three key contributions. 1) We propose a knowledge-enhanced data generation pipeline to provide scalable and high-quality training data for agent development. 2) To mitigate interference among heterogeneous tasks, we introduce a Mixture-of-Experts (MoE) architecture with task-level routing. 3) We develop a Multimodal Reasoning-Augmented Reinforcement Learning approach to enhance the agent's reasoning ability for visual diversity in Minecraft. Built upon these innovations, we present Optimus-3, a general-purpose agent for Minecraft. Extensive experimental results demonstrate that Optimus-3 surpasses both generalist multimodal large language models and existing state-of-the-art agents across a wide range of tasks in the Minecraft environment. Project page: https://cybertronagent.github.io/Optimus-3.github.io/
Authors: Jim O'Connor, Yeonghun Lee, Gary B Parker
Abstract: StarCraft: Brood War remains a challenging benchmark for artificial intelligence research, particularly in the domain of macromanagement, where long-term strategic planning is required. Traditional approaches to StarCraft AI rely on rule-based systems or supervised deep learning, both of which face limitations in adaptability and computational efficiency. In this work, we introduce NeuroPAL, a neuroevolutionary framework that integrates Neuroevolution of Augmenting Topologies (NEAT) with Punctuated Anytime Learning (PAL) to improve the efficiency of evolutionary training. By alternating between frequent, low-fidelity training and periodic, high-fidelity evaluations, PAL enhances the sample efficiency of NEAT, enabling agents to discover effective strategies in fewer training iterations. We evaluate NeuroPAL in a fixed-map, single-race scenario in StarCraft: Brood War and compare its performance to standard NEAT-based training. Our results show that PAL significantly accelerates the learning process, allowing the agent to reach competitive levels of play in approximately half the training time required by NEAT alone. Additionally, the evolved agents exhibit emergent behaviors such as proxy barracks placement and defensive building optimization, strategies commonly used by expert human players. These findings suggest that structured evaluation mechanisms like PAL can enhance the scalability and effectiveness of neuroevolution in complex real-time strategy environments.
Authors: Yuquan Xie, Zaijing Li, Rui Shao, Gongwei Chen, Kaiwen Zhou, Yinchuan Li, Dongmei Jiang, Liqiang Nie
Abstract: Recent efforts to leverage the Multi-modal Large Language Model (MLLM) as GUI agents have yielded promising outcomes. However, these agents still struggle with long-horizon tasks in online environments, primarily due to insufficient knowledge and the inherent gap between offline and online domains. In this paper, inspired by how humans generalize knowledge in open-ended environments, we propose a Hierarchical Multimodal Skills (HMS) module to tackle the issue of insufficient knowledge. It progressively abstracts trajectories into execution skills, core skills, and ultimately meta-skills, providing a hierarchical knowledge structure for long-horizon task planning. To bridge the domain gap, we propose the Skill-Augmented Monte Carlo Tree Search (SA-MCTS) algorithm, which efficiently leverages skills acquired in offline environments to reduce the action search space during online tree exploration. Building on HMS, we propose Mirage-1, a multimodal, cross-platform, plug-and-play GUI agent. To validate the performance of Mirage-1 in real-world long-horizon scenarios, we constructed a new benchmark, AndroidLH. Experimental results show that Mirage-1 outperforms previous agents by 32\%, 19\%, 15\%, and 79\% on AndroidWorld, MobileMiniWob++, Mind2Web-Live, and AndroidLH, respectively. Project page: https://cybertronagent.github.io/Mirage-1.github.io/
Authors: Jintao Liang, Gang Su, Huifeng Lin, You Wu, Rui Zhao, Ziyue Li
Abstract: Retrieval-Augmented Generation (RAG) has emerged as a powerful framework to overcome the knowledge limitations of Large Language Models (LLMs) by integrating external retrieval with language generation. While early RAG systems based on static pipelines have shown effectiveness in well-structured tasks, they struggle in real-world scenarios requiring complex reasoning, dynamic retrieval, and multi-modal integration. To address these challenges, the field has shifted toward Reasoning Agentic RAG, a paradigm that embeds decision-making and adaptive tool use directly into the retrieval process. In this paper, we present a comprehensive review of Reasoning Agentic RAG methods, categorizing them into two primary systems: predefined reasoning, which follows fixed modular pipelines to boost reasoning, and agentic reasoning, where the model autonomously orchestrates tool interaction during inference. We analyze representative techniques under both paradigms, covering architectural design, reasoning strategies, and tool coordination. Finally, we discuss key research challenges and propose future directions to advance the flexibility, robustness, and applicability of reasoning agentic RAG systems. Our collection of the relevant research has been organized into a https://github.com/ByebyeMonica/Reasoning-Agentic-RAG.
URLs: https://github.com/ByebyeMonica/Reasoning-Agentic-RAG.
Authors: Boris Sedlak, Alireza Furutanpey, Zihang Wang, V\'ictor Casamayor Pujol, Schahram Dustdar
Abstract: Edge computing breaks with traditional autoscaling due to strict resource constraints, thus, motivating more flexible scaling behaviors using multiple elasticity dimensions. This work introduces an agent-based autoscaling framework that dynamically adjusts both hardware resources and internal service configurations to maximize requirements fulfillment in constrained environments. We compare four types of scaling agents: Active Inference, Deep Q Network, Analysis of Structural Knowledge, and Deep Active Inference, using two real-world processing services running in parallel: YOLOv8 for visual recognition and OpenCV for QR code detection. Results show all agents achieve acceptable SLO performance with varying convergence patterns. While the Deep Q Network benefits from pre-training, the structural analysis converges quickly, and the deep active inference agent combines theoretical foundations with practical scalability advantages. Our findings provide evidence for the viability of multi-dimensional agent-based autoscaling for edge environments and encourage future work in this research direction.
Authors: Yaoming Zhu, Junxin Wang, Yiyang Li, Lin Qiu, ZongYu Wang, Jun Xu, Xuezhi Cao, Yuhuai Wei, Mingshi Wang, Xunliang Cai, Rong Ma
Abstract: As models become increasingly sophisticated, conventional algorithm benchmarks are increasingly saturated, underscoring the need for more challenging benchmarks to guide future improvements in algorithmic reasoning. This paper introduces OIBench, a high-quality, private, and challenging olympiad-level informatics dataset comprising 250 carefully curated original problems. We detail the construction methodology of the benchmark, ensuring a comprehensive assessment across various programming paradigms and complexities, and we demonstrate its contamination-resistant properties via experiments. We propose Time/Space Completion Curves for finer-grained efficiency analysis and enable direct human-model comparisons through high-level participant evaluations. Our experiments reveal that while open-source models lag behind closed-source counterparts, current SOTA models already outperform most human participants in both correctness and efficiency, while still being suboptimal compared to the canonical solutions. By releasing OIBench as a fully open-source resource (https://huggingface.co/datasets/AGI-Eval/OIBench), we hope this benchmark will contribute to advancing code reasoning capabilities for future LLMs.
Authors: Yuhao Zhou, Yiheng Wang, Xuming He, Ruoyao Xiao, Zhiwei Li, Qiantai Feng, Zijie Guo, Yuejin Yang, Hao Wu, Wenxuan Huang, Jiaqi Wei, Dan Si, Xiuqi Yao, Jia Bu, Haiwen Huang, Tianfan Fu, Shixiang Tang, Ben Fei, Dongzhan Zhou, Fenghua Ling, Yan Lu, Siqi Sun, Chenhui Li, Guanjie Zheng, Jiancheng Lv, Wenlong Zhang, Lei Bai
Abstract: Scientific discoveries increasingly rely on complex multimodal reasoning based on information-intensive scientific data and domain-specific expertise. Empowered by expert-level scientific benchmarks, scientific Multimodal Large Language Models (MLLMs) hold the potential to significantly enhance this discovery process in realistic workflows. However, current scientific benchmarks mostly focus on evaluating the knowledge understanding capabilities of MLLMs, leading to an inadequate assessment of their perception and reasoning abilities. To address this gap, we present the Scientists' First Exam (SFE) benchmark, designed to evaluate the scientific cognitive capacities of MLLMs through three interconnected levels: scientific signal perception, scientific attribute understanding, scientific comparative reasoning. Specifically, SFE comprises 830 expert-verified VQA pairs across three question types, spanning 66 multimodal tasks across five high-value disciplines. Extensive experiments reveal that current state-of-the-art GPT-o3 and InternVL-3 achieve only 34.08% and 26.52% on SFE, highlighting significant room for MLLMs to improve in scientific realms. We hope the insights obtained in SFE will facilitate further developments in AI-enhanced scientific discoveries.
Authors: Yanan Cai, Ahmed Salem, Besmira Nushi, Mark Russinovich
Abstract: We introduce LogiPlan, a novel benchmark designed to evaluate the capabilities of large language models (LLMs) in logical planning and reasoning over complex relational structures. Logical relational reasoning is important for applications that may rely on LLMs to generate and query structured graphs of relations such as network infrastructure, knowledge bases, or business process schema. Our framework allows for dynamic variation of task complexity by controlling the number of objects, relations, and the minimum depth of relational chains, providing a fine-grained assessment of model performance across difficulty levels. LogiPlan encompasses three complementary tasks: (1) Plan Generation, where models must construct valid directed relational graphs meeting specified structural constraints; (2) Consistency Detection, testing models' ability to identify inconsistencies in relational structures; and (3) Comparison Question, evaluating models' capacity to determine the validity of queried relationships within a given graph. Additionally, we assess models' self-correction capabilities by prompting them to verify and refine their initial solutions. We evaluate state-of-the-art models including DeepSeek R1, Gemini 2.0 Pro, Gemini 2 Flash Thinking, GPT-4.5, GPT-4o, Llama 3.1 405B, O3-mini, O1, and Claude 3.7 Sonnet across these tasks, revealing significant performance gaps that correlate with model scale and architecture. Our analysis demonstrates that while recent reasoning-enhanced models show promising results on simpler instances, they struggle with more complex configurations requiring deeper logical planning.
Authors: Mohd Anwar Jamal Faiz
Abstract: This paper introduces the Primender sequence, a novel integer sequence defined by a hybrid rule that combines classical primality with modular digit-based conditions. Specifically, a number n is included in the sequence if it is prime or ends with a prime number of unit digit or any length. In other words, numbers which are primes or have at least one prime suffix. The resulting sequence exhibits a deterministic yet non-trivial structure, blending number-theoretic properties with symbolic patterning. We propose the Primender sequence as a benchmark for evaluating the symbolic reasoning capabilities of Large Language Models (LLMs). The study is motivated by the need for interpretable, rule-based testbeds that can assess an LLM's ability to infer hidden rules, validate mathematical hypotheses, and generalize symbolic logic at scale. A key hypothesis explored is: Whenever a number in the Primender sequence is exactly one more than the largest prime less than or equal to it, the difference between it and the previous number in the sequence is also 1. We design a structured prompt and evaluation framework to test this hypothesis across multiple state-of-the-art LLMs, including ChatGPT, Copilot, DeepSeek, Gemini, Grok, and LLaMA. The models are tasked with identifying the underlying rule, validating the hypothesis, and generating the next 100,000 terms of the sequence. Comparative metrics such as rule inference accuracy, hypothesis evaluation, sequence validity, and symbolic explanation quality are used to assess model performance. This work contributes a novel mathematical construct and a reproducible methodology for benchmarking LLMs in symbolic reasoning, hypothesis testing, and scalable pattern generalization - bridging the domains of number theory, artificial intelligence, and software engineering.
Authors: Henrik Sebastian Steude, Alexander Diedrich, Ingo Pill, Lukas Moddemann, Daniel Vranje\v{s}, Oliver Niggemann
Abstract: Diagnostic processes for complex cyber-physical systems often require extensive prior knowledge in the form of detailed system models or comprehensive training data. However, obtaining such information poses a significant challenge. To address this issue, we present a new diagnostic approach that operates with minimal prior knowledge, requiring only a basic understanding of subsystem relationships and data from nominal operations. Our method combines a neural network-based symptom generator, which employs subsystem-level anomaly detection, with a new graph diagnosis algorithm that leverages minimal causal relationship information between subsystems-information that is typically available in practice. Our experiments with fully controllable simulated datasets show that our method includes the true causal component in its diagnosis set for 82 p.c. of all cases while effectively reducing the search space in 73 p.c. of the scenarios. Additional tests on the real-world Secure Water Treatment dataset showcase the approach's potential for practical scenarios. Our results thus highlight our approach's potential for practical applications with large and complex cyber-physical systems where limited prior knowledge is available.
Authors: Vincenzo Colle, Mohamed Sana, Nicola Piovesan, Antonio De Domenico, Fadhel Ayed, Merouane Debbah
Abstract: The increasing adoption of artificial intelligence in telecommunications has raised interest in the capability of Large Language Models (LLMs) to address domain-specific, mathematically intensive tasks. Although recent advancements have improved the performance of LLMs in general mathematical reasoning, their effectiveness within specialized domains, such as signal processing, network optimization, and performance analysis, remains largely unexplored. To address this gap, we introduce TeleMath, the first benchmark dataset specifically designed to evaluate LLM performance in solving mathematical problems with numerical solutions in the telecommunications domain. Comprising 500 question-answer (QnA) pairs, TeleMath covers a wide spectrum of topics in the telecommunications field. This paper outlines the proposed QnAs generation pipeline, starting from a selected seed of problems crafted by Subject Matter Experts. The evaluation of a wide range of open-source LLMs reveals that best performance on TeleMath is achieved by recent models explicitly designed for mathematical or logical reasoning. In contrast, general-purpose models, even those with a large number of parameters, often struggle with these challenges. We have released the dataset and the evaluation code to ease result reproducibility and support future research.
Authors: Tom Westermann, Aljosha K\"ocher, Felix Gehlhoff
Abstract: AutomationML (AML) enables standardized data exchange in engineering, yet existing recommendations for proper AML modeling are typically formulated as informal and textual constraints. These constraints cannot be validated automatically within AML itself. This work-in-progress paper introduces a pipeline to formalize and verify such constraints. First, AML models are mapped to OWL ontologies via RML and SPARQL. In addition, a Large Language Model translates textual rules into SHACL constraints, which are then validated against the previously generated AML ontology. Finally, SHACL validation results are automatically interpreted in natural language. The approach is demonstrated on a sample AML recommendation. Results show that even complex modeling rules can be semi-automatically checked -- without requiring users to understand formal methods or ontology technologies.
Authors: Michael Bartholomew, Joohyung Lee
Abstract: Answer Set Programming Modulo Theories (ASPMT) is an approach to combining answer set programming and satisfiability modulo theories based on the functional stable model semantics. It is shown that the tight fragment of ASPMT programs can be turned into SMT instances, thereby allowing SMT solvers to compute stable models of ASPMT programs. In this paper we present a compiler called {\sc aspsmt2smt}, which implements this translation. The system uses ASP grounder {\sc gringo} and SMT solver {\sc z3}. {\sc gringo} partially grounds input programs while leaving some variables to be processed by {\sc z3}. We demonstrate that the system can effectively handle real number computations for reasoning about continuous changes.
Authors: Adam Ishay, Zhun Yang, Joohyung Lee, Ilgu Kang, Dongjae Lim
Abstract: Causal and temporal reasoning about video dynamics is a challenging problem. While neuro-symbolic models that combine symbolic reasoning with neural-based perception and prediction have shown promise, they exhibit limitations, especially in answering counterfactual questions. This paper introduces a method to enhance a neuro-symbolic model for counterfactual reasoning, leveraging symbolic reasoning about causal relations among events. We define the notion of a causal graph to represent such relations and use Answer Set Programming (ASP), a declarative logic programming method, to find how to coordinate perception and simulation modules. We validate the effectiveness of our approach on two benchmarks, CLEVRER and CRAFT. Our enhancement achieves state-of-the-art performance on the CLEVRER challenge, significantly outperforming existing models. In the case of the CRAFT benchmark, we leverage a large pre-trained language model, such as GPT-3.5 and GPT-4, as a proxy for a dynamics simulator. Our findings show that this method can further improve its performance on counterfactual questions by providing alternative prompts instructed by symbolic causal reasoning.
Authors: Xiaozhe Li, Jixuan Chen, Xinyu Fang, Shengyuan Ding, Haodong Duan, Qingwen Liu, Kai Chen
Abstract: Large Language Models (LLMs) have shown remarkable capabilities in solving diverse tasks. However, their proficiency in iteratively optimizing complex solutions through learning from previous feedback remains insufficiently explored. To bridge this gap, we present OPT-BENCH, a comprehensive benchmark designed to evaluate LLM agents on large-scale search space optimization problems. OPT-BENCH includes 20 real-world machine learning tasks sourced from Kaggle and 10 classical NP problems, offering a diverse and challenging environment for assessing LLM agents on iterative reasoning and solution refinement. To enable rigorous evaluation, we introduce OPT-Agent, an end-to-end optimization framework that emulates human reasoning when tackling complex problems by generating, validating, and iteratively improving solutions through leveraging historical feedback. Through extensive experiments on 9 state-of-the-art LLMs from 6 model families, we analyze the effects of optimization iterations, temperature settings, and model architectures on solution quality and convergence. Our results demonstrate that incorporating historical context significantly enhances optimization performance across both ML and NP tasks. All datasets, code, and evaluation tools are open-sourced to promote further research in advancing LLM-driven optimization and iterative reasoning. Project page: \href{https://github.com/OliverLeeXZ/OPT-BENCH}{https://github.com/OliverLeeXZ/OPT-BENCH}.
URLs: https://github.com/OliverLeeXZ/OPT-BENCH, https://github.com/OliverLeeXZ/OPT-BENCH
Authors: Yu Zhang, Yang Hu, De Wang
Abstract: Human spatiotemporal behavior simulation is critical for urban planning research, yet traditional rule-based and statistical approaches suffer from high computational costs, limited generalizability, and poor scalability. While large language models (LLMs) show promise as "world simulators," they face challenges in spatiotemporal reasoning including limited spatial cognition, lack of physical constraint understanding, and group homogenization tendencies. This paper introduces a framework integrating chain-of-thought (CoT) reasoning with Model Context Protocol (MCP) to enhance LLMs' capability in simulating spatiotemporal behaviors that correspond with validation data patterns. The methodology combines human-like progressive reasoning through a five-stage cognitive framework with comprehensive data processing via six specialized MCP tool categories: temporal management, spatial navigation, environmental perception, personal memory, social collaboration, and experience evaluation. Experiments in Shanghai's Lujiazui district validate the framework's effectiveness across 1,000 generated samples. Results demonstrate high similarity with real mobile signaling data, achieving generation quality scores of 7.86 to 8.36 across different base models. Parallel processing experiments show efficiency improvements, with generation times decreasing from 1.30 to 0.17 minutes per sample when scaling from 2 to 12 processes. This work contributes to integrating CoT reasoning with MCP for urban behavior modeling, advancing LLMs applications in urban computing and providing a practical approach for synthetic mobility data generation. The framework offers a foundation for smart city planning, transportation forecasting, and participatory urban design applications.
Authors: Daniel Borrajo, Giuseppe Canonaco, Tom\'as de la Rosa, Alfredo Garrach\'on, Sriram Gopalakrishnan, Simerjot Kaur, Marianela Morales, Sunandita Patra, Alberto Pozanco, Keshav Ramani, Charese Smiley, Pietro Totis, Manuela Veloso
Abstract: Classical AI Planning techniques generate sequences of actions for complex tasks. However, they lack the ability to understand planning tasks when provided using natural language. The advent of Large Language Models (LLMs) has introduced novel capabilities in human-computer interaction. In the context of planning tasks, LLMs have shown to be particularly good in interpreting human intents among other uses. This paper introduces GenPlanX that integrates LLMs for natural language-based description of planning tasks, with a classical AI planning engine, alongside an execution and monitoring framework. We demonstrate the efficacy of GenPlanX in assisting users with office-related tasks, highlighting its potential to streamline workflows and enhance productivity through seamless human-AI collaboration.
Authors: Fei Lin, Ziyang Gong, Cong Wang, Yonglin Tian, Tengchao Zhang, Xue Yang, Gen Luo, Fei-Yue Wang
Abstract: Toxicity remains a leading cause of early-stage drug development failure. Despite advances in molecular design and property prediction, the task of molecular toxicity repair - generating structurally valid molecular alternatives with reduced toxicity - has not yet been systematically defined or benchmarked. To fill this gap, we introduce ToxiMol, the first benchmark task for general-purpose Multimodal Large Language Models (MLLMs) focused on molecular toxicity repair. We construct a standardized dataset covering 11 primary tasks and 560 representative toxic molecules spanning diverse mechanisms and granularities. We design a prompt annotation pipeline with mechanism-aware and task-adaptive capabilities, informed by expert toxicological knowledge. In parallel, we propose an automated evaluation framework, ToxiEval, which integrates toxicity endpoint prediction, synthetic accessibility, drug-likeness, and structural similarity into a high-throughput evaluation chain for repair success. We systematically assess nearly 30 mainstream general-purpose MLLMs and design multiple ablation studies to analyze key factors such as evaluation criteria, candidate diversity, and failure attribution. Experimental results show that although current MLLMs still face significant challenges on this task, they begin to demonstrate promising capabilities in toxicity understanding, semantic constraint adherence, and structure-aware molecule editing.
Authors: Rulin Shao, Shuyue Stella Li, Rui Xin, Scott Geng, Yiping Wang, Sewoong Oh, Simon Shaolei Du, Nathan Lambert, Sewon Min, Ranjay Krishna, Yulia Tsvetkov, Hannaneh Hajishirzi, Pang Wei Koh, Luke Zettlemoyer
Abstract: We show that reinforcement learning with verifiable rewards (RLVR) can elicit strong mathematical reasoning in certain models even with spurious rewards that have little, no, or even negative correlation with the correct answer. For example, RLVR improves MATH-500 performance for Qwen2.5-Math-7B in absolute points by 21.4% (random reward), 13.8% (format reward), 24.1% (incorrect label), 26.0% (1-shot RL), and 27.1% (majority voting) -- nearly matching the 29.1% gained with ground truth rewards. However, the spurious rewards that work for Qwen often fail to yield gains with other model families like Llama3 or OLMo2. In particular, we find code reasoning -- thinking in code without actual code execution -- to be a distinctive Qwen2.5-Math behavior that becomes significantly more frequent after RLVR, from 65% to over 90%, even with spurious rewards. Overall, we hypothesize that, given the lack of useful reward signal, RLVR must somehow be surfacing useful reasoning representations learned during pretraining, although the exact mechanism remains a topic for future work. We suggest that future RLVR research should possibly be validated on diverse models rather than a single de facto choice, as we show that it is easy to get significant performance gains on Qwen models even with completely spurious reward signals.
Authors: Shangshang Wang, Julian Asilis, \"Omer Faruk Akg\"ul, Enes Burak Bilgin, Ollie Liu, Willie Neiswanger
Abstract: How cost-effectively can strong reasoning abilities be achieved in language models? Driven by this fundamental question, we present Tina, a family of tiny reasoning models achieved with high cost-efficiency. Notably, Tina demonstrates that substantial reasoning performance can be developed using only minimal resources, by applying parameter-efficient updates during reinforcement learning (RL), using low-rank adaptation (LoRA), to an already tiny 1.5B parameter base model. This minimalist approach produces models that achieve reasoning performance which is competitive with, and sometimes surpasses, SOTA RL reasoning models built upon the same base model. Crucially, this is achieved at a tiny fraction of the computational post-training cost employed by existing SOTA models. In fact, the best Tina model achieves a >20\% reasoning performance increase and 43.33\% Pass@1 accuracy on AIME24, at only \$9 USD post-training and evaluation cost (i.e., an estimated 260x cost reduction). Our work reveals the surprising effectiveness of efficient RL reasoning via LoRA. We validate this across multiple open-source reasoning datasets and various ablation settings starting with a single, fixed set of hyperparameters. Furthermore, we hypothesize that this effectiveness and efficiency stem from LoRA rapidly adapting the model to the structural format of reasoning rewarded by RL, while largely preserving the base model's underlying knowledge. In service of accessibility and open research, we fully open-source all code, training logs, and model weights \& checkpoints.
Authors: Shangshang Wang, Julian Asilis, \"Omer Faruk Akg\"ul, Enes Burak Bilgin, Ollie Liu, Deqing Fu, Willie Neiswanger
Abstract: How cost-effectively can we elicit strong reasoning in language models by leveraging their underlying representations? We answer this question with Resa, a family of 1.5B reasoning models trained via a novel and efficient sparse autoencoder tuning (SAE-Tuning) procedure. This method first trains an SAE to capture reasoning abilities from a source model, and then uses the trained SAE to guide a standard supervised fine-tuning process to elicit such abilities in a target model, all using verified question-answer data without any reasoning traces. Notably, when applied to certain base models before further RL post-training, SAE-Tuning retains >97% of its RL-trained counterpart's reasoning performance while reducing training costs by >2000x to roughly \$1 and training time by >450x to around 20 minutes. Furthermore, when applied to lightly RL-trained models (e.g., within 1 hour on 2 GPUs), it enables reasoning performance such as 43.33% Pass@1 on AIME24 and 90% Pass@1 on AMC23 for only around \$1 additional cost. Surprisingly, the reasoning abilities extracted via SAEs are potentially both generalizable and modular. Generality means abilities extracted from one dataset still elevate performance on a larger and overlapping corpus. Modularity means abilities extracted from Qwen or Qwen-Math can be attached to the R1-Distill model at test time, without any retraining, and yield comparable gains. Extensive ablations validate these findings and all artifacts are fully open-sourced.
Authors: Yuxuan Li, Sheng Jinag, Bizhu Wang
Abstract: With technology advancing and the pursuit of new audiovisual experiences strengthening, the metaverse has gained surging enthusiasm. However, it faces practical hurdles as substantial data like high-resolution virtual scenes must be transmitted between cloud platforms and VR devices. Specifically, the VR device's wireless transmission hampered by insufficient bandwidth, causes speed and delay problems. Meanwhile, poor channel quality leads to data errors and worsens user experience. To solve this, we've proposed the Semantic Communication-Enabled Cloud-Edge-End Collaborative Immersive Metaverse Service (SC-CEE-Meta) Architecture, which includes three modules: VR video semantic transmission, video synthesis, and 3D virtual scene reconstruction. By deploying semantic modules on VR devices and edge servers and sending key semantic info instead of focusing on bit-level reconstruction, it can cut latency, resolve the resource-bandwidth conflict, and better withstand channel interference. Also, the cloud deploys video synthesis and 3D scene reconstruction preprocessing, while edge devices host 3D reconstruction rendering modules, all for immersive services. Verified on Meta Quest Pro, the SC-CEE-Meta can reduce wireless transmission delay by 96.05\% and boost image quality by 43.99\% under poor channel condition.
Authors: Jianwu Fang, Lei-Lei Li, Zhedong Zheng, Hongkai Yu, Jianru Xue, Zhengguo Li, Tat-Seng Chua
Abstract: Traffic Accident Anticipation (TAA) in traffic scenes is a challenging problem for achieving zero fatalities in the future. Current approaches typically treat TAA as a supervised learning task needing the laborious annotation of accident occurrence duration. However, the inherent long-tailed, uncertain, and fast-evolving nature of traffic scenes has the problem that real causal parts of accidents are difficult to identify and are easily dominated by data bias, resulting in a background confounding issue. Thus, we propose an Attentive Video Diffusion (AVD) model that synthesizes additional accident video clips by generating the causal part in dashcam videos, i.e., from normal clips to accident clips. AVD aims to generate causal video frames based on accident or accident-free text prompts while preserving the style and content of frames for TAA after video generation. This approach can be trained using datasets collected from various driving scenes without any extra annotations. Additionally, AVD facilitates an Equivariant TAA (EQ-TAA) with an equivariant triple loss for an anchor accident-free video clip, along with the generated pair of contrastive pseudo-normal and pseudo-accident clips. Extensive experiments have been conducted to evaluate the performance of AVD and EQ-TAA, and competitive performance compared to state-of-the-art methods has been obtained.
Authors: Haopeng Wang, Haiwei Dong, Abdulmotaleb El Saddik
Abstract: Extended reality (XR) is rapidly advancing, and poised to revolutionize content creation and consumption. In XR, users integrate various sensory inputs to form a cohesive perception of the virtual environment. This survey reviews the state-of-the-art in XR streaming, focusing on multiple paradigms. To begin, we define XR and introduce various XR headsets along with their multimodal interaction methods to provide a foundational understanding. We then analyze XR traffic characteristics to highlight the unique data transmission requirements. We also explore factors that influence the quality of experience in XR systems, aiming to identify key elements for enhancing user satisfaction. Following this, we present visual attention-based optimization methods for XR streaming to improve efficiency and performance. Finally, we examine current applications and highlight challenges to provide insights into ongoing and future developments of XR.
Authors: Sridhar S, Nithin A, Shakeel Rifath, Vasantha Raj
Abstract: Advances in generative artificial intelligence have altered multimedia creation, allowing for automatic cinematic video synthesis from text inputs. This work describes a method for creating 60-second cinematic movies incorporating Stable Diffusion for high-fidelity image synthesis, GPT-2 for narrative structuring, and a hybrid audio pipeline using gTTS and YouTube-sourced music. It uses a five-scene framework, which is augmented by linear frame interpolation, cinematic post-processing (e.g., sharpening), and audio-video synchronization to provide professional-quality results. It was created in a GPU-accelerated Google Colab environment using Python 3.11. It has a dual-mode Gradio interface (Simple and Advanced), which supports resolutions of up to 1024x768 and frame rates of 15-30 FPS. Optimizations such as CUDA memory management and error handling ensure reliability. The experiments demonstrate outstanding visual quality, narrative coherence, and efficiency, furthering text-to-video synthesis for creative, educational, and industrial applications.
Authors: Jie Qin, Wei Yang, Yan Su, Yiran Zhu, Weizhen Li, Yunyue Pan, Chengchang Pan, Honggang Qi
Abstract: Current HER2 assessment models for breast cancer predominantly analyze H&E or IHC images in isolation,despite clinical reliance on their synergistic interpretation. However, concurrent acquisition of both modalities is often hindered by workflow complexity and cost constraints. We propose an adaptive bimodal framework enabling flexible single-/dual-modality HER2 prediction through three innovations: 1) A dynamic branch selector that activates either single-modality reconstruction or dual-modality joint inference based on input completeness; 2) A bidirectional cross-modal GAN performing context-aware feature-space reconstruction of missing modalities; 3) A hybrid training protocol integrating adversarial learning and multi-task optimization. This architecture elevates single-modality H&E prediction accuracy from 71.44% to 94.25% while achieving 95.09% dual-modality accuracy, maintaining 90.28% reliability with sole IHC inputs. The framework's "dual-preferred, single-compatible" design delivers near-bimodal performance without requiring synchronized acquisition, particularly benefiting resource-limited settings through IHC infrastructure cost reduction. Experimental validation confirms 22.81%/12.90% accuracy improvements over H&E/IHC baselines respectively, with cross-modal reconstruction enhancing F1-scores to 0.9609 (HE to IHC) and 0.9251 (IHC to HE). By dynamically routing inputs through reconstruction-enhanced or native fusion pathways, the system mitigates performance degradation from missing data while preserving computational efficiency (78.55% parameter reduction in lightweight variant). This elastic architecture demonstrates significant potential for democratizing precise HER2 assessment across diverse healthcare settings.
Authors: Kangwei Liu, Junwu Liu, Xiaowei Yi, Jinlin Guo, Yun Cao
Abstract: Audio-driven emotional 3D facial animation encounters two significant challenges: (1) reliance on single-modal control signals (videos, text, or emotion labels) without leveraging their complementary strengths for comprehensive emotion manipulation, and (2) deterministic regression-based mapping that constrains the stochastic nature of emotional expressions and non-verbal behaviors, limiting the expressiveness of synthesized animations. To address these challenges, we present a diffusion-based framework for controllable expressive 3D facial animation. Our approach introduces two key innovations: (1) a FLAME-centered multimodal emotion binding strategy that aligns diverse modalities (text, audio, and emotion labels) through contrastive learning, enabling flexible emotion control from multiple signal sources, and (2) an attention-based latent diffusion model with content-aware attention and emotion-guided layers, which enriches motion diversity while maintaining temporal coherence and natural facial dynamics. Extensive experiments demonstrate that our method outperforms existing approaches across most metrics, achieving a 21.6\% improvement in emotion similarity while preserving physiologically plausible facial dynamics. Project Page: https://kangweiiliu.github.io/Control_3D_Animation.
Authors: Yi-Chun Chen
Abstract: This paper presents a hierarchical knowledge graph framework for the structured understanding of visual narratives, focusing on multimodal media such as comics. The proposed method decomposes narrative content into multiple levels, from macro-level story arcs to fine-grained event segments. It represents them through integrated knowledge graphs that capture semantic, spatial, and temporal relationships. At the panel level, we construct multimodal graphs that link visual elements such as characters, objects, and actions with corresponding textual components, including dialogue and captions. These graphs are integrated across narrative levels to support reasoning over story structure, character continuity, and event progression. We apply our approach to a manually annotated subset of the Manga109 dataset and demonstrate its ability to support symbolic reasoning across diverse narrative tasks, including action retrieval, dialogue tracing, character appearance mapping, and panel timeline reconstruction. Evaluation results show high precision and recall across tasks, validating the coherence and interpretability of the framework. This work contributes a scalable foundation for narrative-based content analysis, interactive storytelling, and multimodal reasoning in visual media.
Authors: Weiyin Gong, Kai Zhang, Yanghai Zhang, Qi Liu, Xinjie Sun, Junyu Lu, Linbo Zhu
Abstract: Multimodal intent recognition (MIR) seeks to accurately interpret user intentions by integrating verbal and non-verbal information across video, audio and text modalities. While existing approaches prioritize text analysis, they often overlook the rich semantic content embedded in non-verbal cues. This paper presents a novel Wavelet-Driven Multimodal Intent Recognition(WDMIR) framework that enhances intent understanding through frequency-domain analysis of non-verbal information. To be more specific, we propose: (1) a wavelet-driven fusion module that performs synchronized decomposition and integration of video-audio features in the frequency domain, enabling fine-grained analysis of temporal dynamics; (2) a cross-modal interaction mechanism that facilitates progressive feature enhancement from bimodal to trimodal integration, effectively bridging the semantic gap between verbal and non-verbal information. Extensive experiments on MIntRec demonstrate that our approach achieves state-of-the-art performance, surpassing previous methods by 1.13% on accuracy. Ablation studies further verify that the wavelet-driven fusion module significantly improves the extraction of semantic information from non-verbal sources, with a 0.41% increase in recognition accuracy when analyzing subtle emotional cues.
Authors: Longzhen Han, Awes Mubarak, Almas Baimagambetov, Nikolaos Polatidis, Thar Baker
Abstract: Multimodal Large Language Models (MLLMs) have rapidly evolved beyond text generation, now spanning diverse output modalities including images, music, video, human motion, and 3D objects, by integrating language with other sensory modalities under unified architectures. This survey categorises six primary generative modalities and examines how foundational techniques, namely Self-Supervised Learning (SSL), Mixture of Experts (MoE), Reinforcement Learning from Human Feedback (RLHF), and Chain-of-Thought (CoT) prompting, enable cross-modal capabilities. We analyze key models, architectural trends, and emergent cross-modal synergies, while highlighting transferable techniques and unresolved challenges. Architectural innovations like transformers and diffusion models underpin this convergence, enabling cross-modal transfer and modular specialization. We highlight emerging patterns of synergy, and identify open challenges in evaluation, modularity, and structured reasoning. This survey offers a unified perspective on MLLM development and identifies critical paths toward more general-purpose, adaptive, and interpretable multimodal systems.
Authors: Tian Lan, Yang-Hao Zhou, Zi-Ao Ma, Fanshu Sun, Rui-Qing Sun, Junyu Luo, Rong-Cheng Tu, Heyan Huang, Chen Xu, Zhijing Wu, Xian-Ling Mao
Abstract: Recent advances in deep learning have significantly enhanced generative AI capabilities across text, images, and audio. However, automatically evaluating the quality of these generated outputs presents ongoing challenges. Although numerous automatic evaluation methods exist, current research lacks a systematic framework that comprehensively organizes these methods across text, visual, and audio modalities. To address this issue, we present a comprehensive review and a unified taxonomy of automatic evaluation methods for generated content across all three modalities; We identify five fundamental paradigms that characterize existing evaluation approaches across these domains. Our analysis begins by examining evaluation methods for text generation, where techniques are most mature. We then extend this framework to image and audio generation, demonstrating its broad applicability. Finally, we discuss promising directions for future research in cross-modal evaluation methodologies.
Authors: Kyubyung Chae, Hyunbin Jin, Taesup Kim
Abstract: Safely aligning large language models (LLMs) often demands extensive human-labeled preference data, a process that's both costly and time-consuming. While synthetic data offers a promising alternative, current methods frequently rely on complex iterative prompting or auxiliary models. To address this, we introduce Refusal-Aware Adaptive Injection (RAAI), a straightforward, training-free, and model-agnostic framework that repurposes LLM attack techniques. RAAI works by detecting internal refusal signals and adaptively injecting predefined phrases to elicit harmful, yet fluent, completions. Our experiments show RAAI effectively jailbreaks LLMs, increasing the harmful response rate from a baseline of 2.15% to up to 61.04% on average across four benchmarks. Crucially, fine-tuning LLMs with the synthetic data generated by RAAI improves model robustness against harmful prompts while preserving general capabilities on standard tasks like MMLU and ARC. This work highlights how LLM attack methodologies can be reframed as practical tools for scalable and controllable safety alignment.
Authors: Jordi de la Torre
Abstract: We propose a novel architecture for integrating large language models (LLMs) with a persistent, interactive Lisp environment. This setup enables LLMs to define, invoke, and evolve their own tools through programmatic interaction with a live REPL. By embedding Lisp expressions within generation and intercepting them via a middleware layer, the system allows for stateful external memory, reflective programming, and dynamic tool creation. We present a design framework and architectural principles to guide future implementations of interactive AI systems that integrate symbolic programming with neural language generation.
Authors: Haoyang Li, Huan Gao, Zhiyuan Zhao, Zhiyu Lin, Junyu Gao, Xuelong Li
Abstract: The widespread adoption of Large Language Models (LLMs) has heightened concerns about their security, particularly their vulnerability to jailbreak attacks that leverage crafted prompts to generate malicious outputs. While prior research has been conducted on general security capabilities of LLMs, their specific susceptibility to jailbreak attacks in code generation remains largely unexplored. To fill this gap, we propose MalwareBench, a benchmark dataset containing 3,520 jailbreaking prompts for malicious code-generation, designed to evaluate LLM robustness against such threats. MalwareBench is based on 320 manually crafted malicious code generation requirements, covering 11 jailbreak methods and 29 code functionality categories. Experiments show that mainstream LLMs exhibit limited ability to reject malicious code-generation requirements, and the combination of multiple jailbreak methods further reduces the model's security capabilities: specifically, the average rejection rate for malicious content is 60.93%, dropping to 39.92% when combined with jailbreak attack algorithms. Our work highlights that the code security capabilities of LLMs still pose significant challenges.
Authors: Elena Sofia Ruzzetti, Giancarlo A. Xompero, Davide Venditti, Fabio Massimo Zanzotto
Abstract: Large Language Models (LLMs) memorize, and thus, among huge amounts of uncontrolled data, may memorize Personally Identifiable Information (PII), which should not be stored and, consequently, not leaked. In this paper, we introduce Private Memorization Editing (PME), an approach for preventing private data leakage that turns an apparent limitation, that is, the LLMs' memorization ability, into a powerful privacy defense strategy. While attacks against LLMs have been performed exploiting previous knowledge regarding their training data, our approach aims to exploit the same kind of knowledge in order to make a model more robust. We detect a memorized PII and then mitigate the memorization of PII by editing a model knowledge of its training data. We verify that our procedure does not affect the underlying language model while making it more robust against privacy Training Data Extraction attacks. We demonstrate that PME can effectively reduce the number of leaked PII in a number of configurations, in some cases even reducing the accuracy of the privacy attacks to zero.
Authors: Rafa\"el Nouailles (GdR)
Abstract: Large Language models (LLMs) are transforming digital usage, particularly in text generation, image creation, information retrieval and code development. ChatGPT, launched by OpenAI in November 2022, quickly became a reference, prompting the emergence of competitors such as Google's Gemini. However, these technological advances raise new cybersecurity challenges, including prompt injection attacks, the circumvention of regulatory measures (jailbreaking), the spread of misinformation (hallucinations) and risks associated with deep fakes. This paper presents a comparative analysis of the security and alignment levels of ChatGPT and Gemini, as well as a taxonomy of jailbreak techniques associated with experiments.
Authors: Tianyu Chen, Jian Lou, Wenjie Wang
Abstract: As Retrieval-Augmented Generation (RAG) evolves into service-oriented platforms (Rag-as-a-Service) with shared knowledge bases, protecting the copyright of contributed data becomes essential. Existing watermarking methods in RAG focus solely on textual knowledge, leaving image knowledge unprotected. In this work, we propose AQUA, the first watermark framework for image knowledge protection in Multimodal RAG systems. AQUA embeds semantic signals into synthetic images using two complementary methods: acronym-based triggers and spatial relationship cues. These techniques ensure watermark signals survive indirect watermark propagation from image retriever to textual generator, being efficient, effective and imperceptible. Experiments across diverse models and datasets show that AQUA enables robust, stealthy, and reliable copyright tracing, filling a key gap in multimodal RAG protection.
Authors: Fuhan Cai, Yong Guo, Jie Li, Wenbo Li, Xiangzhong Fang, Jian Chen
Abstract: Recent advancements in text-to-image (T2I) generation have led to the emergence of highly expressive models such as diffusion transformers (DiTs), exemplified by FLUX. However, their massive parameter sizes lead to slow inference, high memory usage, and poor deployability. Existing acceleration methods (e.g., single-step distillation and attention pruning) often suffer from significant performance degradation and incur substantial training costs. To address these limitations, we propose FastFLUX, an architecture-level pruning framework designed to enhance the inference efficiency of FLUX. At its core is the Block-wise Replacement with Linear Layers (BRLL) method, which replaces structurally complex residual branches in ResBlocks with lightweight linear layers while preserving the original shortcut connections for stability. Furthermore, we introduce Sandwich Training (ST), a localized fine-tuning strategy that leverages LoRA to supervise neighboring blocks, mitigating performance drops caused by structural replacement. Experiments show that our FastFLUX maintains high image quality under both qualitative and quantitative evaluations, while significantly improving inference speed, even with 20\% of the hierarchy pruned. Our code will be available soon.
Authors: Giannis Daras, Adrian Rodriguez-Munoz, Adam Klivans, Antonio Torralba, Constantinos Daskalakis
Abstract: We show how to use low-quality, synthetic, and out-of-distribution images to improve the quality of a diffusion model. Typically, diffusion models are trained on curated datasets that emerge from highly filtered data pools from the Web and other sources. We show that there is immense value in the lower-quality images that are often discarded. We present Ambient Diffusion Omni, a simple, principled framework to train diffusion models that can extract signal from all available images during training. Our framework exploits two properties of natural images -- spectral power law decay and locality. We first validate our framework by successfully training diffusion models with images synthetically corrupted by Gaussian blur, JPEG compression, and motion blur. We then use our framework to achieve state-of-the-art ImageNet FID, and we show significant improvements in both image quality and diversity for text-to-image generative modeling. The core insight is that noise dampens the initial skew between the desired high-quality distribution and the mixed distribution we actually observe. We provide rigorous theoretical justification for our approach by analyzing the trade-off between learning from biased data versus limited unbiased data across diffusion times.
Authors: Shangpin Peng, Weinong Wang, Zhuotao Tian, Senqiao Yang, Xing Wu, Haotian Xu, Chengquan Zhang, Takashi Isobe, Baotian Hu, Min Zhang
Abstract: Direct Preference Optimization (DPO) has become a cornerstone of reinforcement learning from human feedback (RLHF) due to its simplicity and efficiency. However, existing DPO-based approaches typically treat all preference pairs uniformly, ignoring critical variations in their inherent quality and learning utility, leading to suboptimal data utilization and performance. To address this challenge, we propose Omni-DPO, a dual-perspective optimization framework that jointly accounts for (1) the inherent quality of each preference pair and (2) the model's evolving performance on those pairs. By adaptively weighting samples according to both data quality and the model's learning dynamics during training, Omni-DPO enables more effective training data utilization and achieves better performance. Experimental results on various models and benchmarks demonstrate the superiority and generalization capabilities of Omni-DPO. On textual understanding tasks, Gemma-2-9b-it finetuned with Omni-DPO beats the leading LLM, Claude 3 Opus, by a significant margin of 6.7 points on the Arena-Hard benchmark. On mathematical reasoning tasks, Omni-DPO consistently outperforms the baseline methods across all benchmarks, providing strong empirical evidence for the effectiveness and robustness of our approach. Code and models will be available at https://github.com/pspdada/Omni-DPO.
Authors: Brendan Leigh Ross, No\"el Vouitsis, Atiyeh Ashari Ghomi, Rasa Hosseinzadeh, Ji Xin, Zhaoyan Liu, Yi Sui, Shiyi Hou, Kin Kwan Leung, Gabriel Loaiza-Ganem, Jesse C. Cresswell
Abstract: Although large language models (LLMs) are becoming increasingly capable of solving challenging real-world tasks, accurately quantifying their uncertainty remains a critical open problem, which limits their applicability in high-stakes domains. This challenge is further compounded by the closed-source, black-box nature of many state-of-the-art LLMs. Moreover, LLM-based systems can be highly sensitive to the prompts that bind them together, which often require significant manual tuning (i.e., prompt engineering). In this work, we address these challenges by viewing LLM-based systems through a Bayesian lens. We interpret prompts as textual parameters in a statistical model, allowing us to use a small training dataset to perform Bayesian inference over these prompts. This novel perspective enables principled uncertainty quantification over both the model's textual parameters and its downstream predictions, while also incorporating prior beliefs about these parameters expressed in free-form text. To perform Bayesian inference, a difficult problem even for well-studied data modalities, we introduce Metropolis-Hastings through LLM Proposals (MHLP), a novel Markov chain Monte Carlo (MCMC) algorithm that combines prompt optimization techniques with standard MCMC methods. MHLP is a turnkey modification to existing LLM pipelines, including those that rely exclusively on closed-source models. Empirically, we demonstrate that our method yields improvements in both predictive accuracy and uncertainty quantification (UQ) on a range of LLM benchmarks and UQ tasks. More broadly, our work demonstrates a viable path for incorporating methods from the rich Bayesian literature into the era of LLMs, paving the way for more reliable and calibrated LLM-based systems.
Authors: Christopher J. Agostino, Quan Le Thien, Molly Apsel, Denizhan Pak, Elina Lesyk, Ashabari Majumdar
Abstract: Semantic degeneracy represents a fundamental property of natural language that extends beyond simple polysemy to encompass the combinatorial explosion of potential interpretations that emerges as semantic expressions increase in complexity. Large Language Models (LLMs) and other modern NLP systems face inherent limitations precisely because they operate within natural language itself, making them subject to the same interpretive constraints imposed by semantic degeneracy. In this work, we argue using Kolmogorov complexity that as an expression's complexity grows, the likelihood of any interpreting agent (human or LLM-powered AI) recovering the single intended meaning vanishes. This computational intractability suggests the classical view that linguistic forms possess meaning in and of themselves is flawed. We alternatively posit that meaning is instead actualized through an observer-dependent interpretive act. To test this, we conducted a semantic Bell inequality test using diverse LLM agents as ``computational cognitive systems'' to interpret ambiguous word pairs under varied contextual settings. Across several independent experiments, we found average CHSH expectation values ranging from 1.2 to 2.8, with several runs yielding values (e.g., 2.3-2.4) that significantly violate the classical boundary ($|S|\leq2$). This demonstrates that linguistic interpretation under ambiguity can exhibit non-classical contextuality, consistent with results from human cognition experiments. These results inherently imply that classical frequentist-based analytical approaches for natural language are necessarily lossy. Instead, we propose that Bayesian-style repeated sampling approaches can provide more practically useful and appropriate characterizations of linguistic meaning in context.
Authors: Christos Ziakas, Alessandra Russo
Abstract: We propose a test-time adaptation method that enables a progress estimation model to adapt online to the visual and temporal context of test trajectories by optimizing a learned self-supervised objective. To this end, we introduce a gradient-based meta-learning strategy to train the model on expert visual trajectories and their natural language task descriptions, such that test-time adaptation improves progress estimation relying on semantic content over temporal order. Our test-time adaptation method generalizes from a single training environment to diverse out-of-distribution tasks, environments, and embodiments, outperforming the state-of-the-art in-context learning approach using autoregressive vision-language models.
Authors: Marcos Abel Zuzu\'arregui, Stefano Carpin
Abstract: Robotics and artificial intelligence hold significant potential for advancing precision agriculture. While robotic systems have been successfully deployed for various tasks, adapting them to perform diverse missions remains challenging, particularly because end users often lack technical expertise. In this paper, we present an end-to-end system that leverages large language models (LLMs), specifically ChatGPT, to enable users to assign complex data collection tasks to autonomous robots using natural language instructions. To enhance reusability, mission plans are encoded using an existing IEEE task specification standard, and are executed on robots via ROS2 nodes that bridge high-level mission descriptions with existing ROS libraries. Through extensive experiments, we highlight the strengths and limitations of LLMs in this context, particularly regarding spatial reasoning and solving complex routing challenges, and show how our proposed implementation overcomes them.
Authors: Ahmed Elbakary, Chaouki Ben Issaid, Mehdi Bennis
Abstract: We present a novel federated multi-task learning method that leverages cross-client similarity to enable personalized learning for each client. To avoid transmitting the entire model to the parameter server, we propose a communication-efficient scheme that introduces a feature anchor, a compact vector representation that summarizes the features learned from the client's local classes. This feature anchor is shared with the server to account for local clients' distribution. In addition, the clients share the classification heads, a lightweight linear layer, and perform a graph-based regularization to enable collaboration among clients. By modeling collaboration between clients as a dynamic graph and continuously updating and refining this graph, we can account for any drift from the clients. To ensure beneficial knowledge transfer and prevent negative collaboration, we leverage a community detection-based approach that partitions this dynamic graph into homogeneous communities, maximizing the sum of task similarities, represented as the graph edges' weights, within each community. This mechanism restricts collaboration to highly similar clients within their formed communities, ensuring positive interaction and preserving personalization. Extensive experiments on two heterogeneous datasets demonstrate that our method significantly outperforms state-of-the-art baselines. Furthermore, we show that our method exhibits superior computation and communication efficiency and promotes fairness across clients.
Authors: Marcos Abel Zuzu\'arregui, Mustafa Melih Toslak, Stefano Carpin
Abstract: Artificial intelligence is transforming precision agriculture, offering farmers new tools to streamline their daily operations. While these technological advances promise increased efficiency, they often introduce additional complexity and steep learning curves that are particularly challenging for non-technical users who must balance tech adoption with existing workloads. In this paper, we present a natural language (NL) robotic mission planner that enables non-specialists to control heterogeneous robots through a common interface. By leveraging large language models (LLMs) and predefined primitives, our architecture seamlessly translates human language into intermediate descriptions that can be executed by different robotic platforms. With this system, users can formulate complex agricultural missions without writing any code. In the work presented in this paper, we extend our previous system tailored for wheeled robot mission planning through a new class of experiments involving robotic manipulation and computer vision tasks. Our results demonstrate that the architecture is both general enough to support a diverse set of robots and powerful enough to execute complex mission requests. This work represents a significant step toward making robotic automation in precision agriculture more accessible to non-technical users.
Authors: Natanael Lucena, F\'abio S. da Silva, Ricardo Rios
Abstract: This paper presents a comparison of the performance of Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) in the task of multi-classifying images containing lesions of psoriasis and diseases similar to it. Models pre-trained on ImageNet were adapted to a specific data set. Both achieved high predictive metrics, but the ViTs stood out for their superior performance with smaller models. Dual Attention Vision Transformer-Base (DaViT-B) obtained the best results, with an f1-score of 96.4%, and is recommended as the most efficient architecture for automated psoriasis detection. This article reinforces the potential of ViTs for medical image classification tasks.
Authors: Maryam Khalid, Akane Sano
Abstract: Graph-based Active Learning (AL) leverages the structure of graphs to efficiently prioritize label queries, reducing labeling costs and user burden in applications like health monitoring, human behavior analysis, and sensor networks. By identifying strategically positioned nodes, graph AL minimizes data collection demands while maintaining model performance, making it a valuable tool for dynamic environments. Despite its potential, existing graph AL methods are often evaluated on static graph datasets and primarily focus on prediction accuracy, neglecting user-centric considerations such as sampling diversity, query fairness, and adaptability to dynamic settings. To bridge this gap, we introduce GRAIL, a novel benchmarking framework designed to evaluate graph AL strategies in dynamic, real-world environments. GRAIL introduces novel metrics to assess sustained effectiveness, diversity, and user burden, enabling a comprehensive evaluation of AL methods under varying conditions. Extensive experiments on datasets featuring dynamic, real-life human sensor data reveal trade-offs between prediction performance and user burden, highlighting limitations in existing AL strategies. GRAIL demonstrates the importance of balancing node importance, query diversity, and network topology, providing an evaluation mechanism for graph AL solutions in dynamic environments.
Authors: Daniel Lawson, Adriana Hugessen, Charlotte Cloutier, Glen Berseth, Khimya Khetarpal
Abstract: Behavioral cloning (BC) methods trained with supervised learning (SL) are an effective way to learn policies from human demonstrations in domains like robotics. Goal-conditioning these policies enables a single generalist policy to capture diverse behaviors contained within an offline dataset. While goal-conditioned behavior cloning (GCBC) methods can perform well on in-distribution training tasks, they do not necessarily generalize zero-shot to tasks that require conditioning on novel state-goal pairs, i.e. combinatorial generalization. In part, this limitation can be attributed to a lack of temporal consistency in the state representation learned by BC; if temporally related states are encoded to similar latent representations, then the out-of-distribution gap for novel state-goal pairs would be reduced. Hence, encouraging this temporal consistency in the representation space should facilitate combinatorial generalization. Successor representations, which encode the distribution of future states visited from the current state, nicely encapsulate this property. However, previous methods for learning successor representations have relied on contrastive samples, temporal-difference (TD) learning, or both. In this work, we propose a simple yet effective representation learning objective, $\text{BYOL-}\gamma$ augmented GCBC, which is not only able to theoretically approximate the successor representation in the finite MDP case without contrastive samples or TD learning, but also, results in competitive empirical performance across a suite of challenging tasks requiring combinatorial generalization.
Authors: Mohammad Taufeeque, Aaron David Tucker, Adam Gleave, Adri\`a Garriga-Alonso
Abstract: We partially reverse-engineer a convolutional recurrent neural network (RNN) trained to play the puzzle game Sokoban with model-free reinforcement learning. Prior work found that this network solves more levels with more test-time compute. Our analysis reveals several mechanisms analogous to components of classic bidirectional search. For each square, the RNN represents its plan in the activations of channels associated with specific directions. These state-action activations are analogous to a value function - their magnitudes determine when to backtrack and which plan branch survives pruning. Specialized kernels extend these activations (containing plan and value) forward and backward to create paths, forming a transition model. The algorithm is also unlike classical search in some ways. State representation is not unified; instead, the network considers each box separately. Each layer has its own plan representation and value function, increasing search depth. Far from being inscrutable, the mechanisms leveraging test-time compute learned in this network by model-free training can be understood in familiar terms.
Authors: Jiaxin Wen, Zachary Ankner, Arushi Somani, Peter Hase, Samuel Marks, Jacob Goldman-Wetzler, Linda Petrini, Henry Sleight, Collin Burns, He He, Shi Feng, Ethan Perez, Jan Leike
Abstract: To steer pretrained language models for downstream tasks, today's post-training paradigm relies on humans to specify desired behaviors. However, for models with superhuman capabilities, it is difficult or impossible to get high-quality human supervision. To address this challenge, we introduce a new unsupervised algorithm, Internal Coherence Maximization (ICM), to fine-tune pretrained language models on their own generated labels, \emph{without external supervision}. On GSM8k-verification, TruthfulQA, and Alpaca reward modeling tasks, our method matches the performance of training on golden supervision and outperforms training on crowdsourced human supervision. On tasks where LMs' capabilities are strongly superhuman, our method can elicit those capabilities significantly better than training on human labels. Finally, we show that our method can improve the training of frontier LMs: we use our method to train an unsupervised reward model and use reinforcement learning to train a Claude 3.5 Haiku-based assistant. Both the reward model and the assistant outperform their human-supervised counterparts.
Authors: Elizabeth Demers, Victor Xiaoqi Wang, Kean Wu
Abstract: Human capital (HC) is increasingly important to corporate value creation. Unlike other assets, however, HC is not currently subject to well-defined measurement or disclosure rules. We use a machine learning algorithm (word2vec) trained on a confirmed set of HC disclosures to develop a comprehensive list of HC-related keywords classified into five subcategories (DEI; health and safety; labor relations and culture; compensation and benefits; and demographics and other) that capture the multidimensional nature of HC management. We share our lexicon, corporate HC disclosures, and the Python code used to develop the lexicon, and we provide detailed examples of using our data and code, including for fine-tuning a BERT model. Researchers can use our HC lexicon (or modify the code to capture another construct of interest) with their samples of corporate communications to address pertinent HC questions. We close with a discussion of future research opportunities related to HC management and disclosure.
Authors: Yi Wang, Max Kreminski
Abstract: Story generation has been a prominent application of Large Language Models (LLMs). However, understanding LLMs' ability to produce high-quality stories remains limited due to challenges in automatic evaluation methods and the high cost and subjectivity of manual evaluation. Computational narratology offers valuable insights into what constitutes a good story, which has been applied in the symbolic narrative planning approach to story generation. This work aims to deepen the understanding of LLMs' story generation capabilities by using them to solve narrative planning problems. We present a benchmark for evaluating LLMs on narrative planning based on literature examples, focusing on causal soundness, character intentionality, and dramatic conflict. Our experiments show that GPT-4 tier LLMs can generate causally sound stories at small scales, but planning with character intentionality and dramatic conflict remains challenging, requiring LLMs trained with reinforcement learning for complex reasoning. The results offer insights on the scale of stories that LLMs can generate while maintaining quality from different aspects. Our findings also highlight interesting problem solving behaviors and shed lights on challenges and considerations for applying LLM narrative planning in game environments.
Authors: Saswat Das, Jameson Sandler, Ferdinando Fioretto
Abstract: Large Language Model agents have begun to appear as personal assistants, customer service bots, and clinical aides. While these applications deliver substantial operational benefits, they also require continuous access to sensitive data, which increases the likelihood of unauthorized disclosures. This study proposes an auditing framework for conversational privacy that quantifies and audits these risks. The proposed Conversational Manipulation for Privacy Leakage (CMPL) framework, is an iterative probing strategy designed to stress-test agents that enforce strict privacy directives. Rather than focusing solely on a single disclosure event, CMPL simulates realistic multi-turn interactions to systematically uncover latent vulnerabilities. Our evaluation on diverse domains, data modalities, and safety configurations demonstrate the auditing framework's ability to reveal privacy risks that are not deterred by existing single-turn defenses. In addition to introducing CMPL as a diagnostic tool, the paper delivers (1) an auditing procedure grounded in quantifiable risk metrics and (2) an open benchmark for evaluation of conversational privacy across agent implementations.
Authors: Yicheng Duan, Kaiyu tang
Abstract: Vision-and-Language Navigation (VLN) presents a complex challenge in embodied AI, requiring agents to interpret natural language instructions and navigate through visually rich, unfamiliar environments. Recent advances in large vision-language models (LVLMs), such as CLIP and Flamingo, have significantly improved multimodal understanding but introduced new challenges related to computational cost and real-time deployment. In this project, we propose a modular, plug-and-play navigation framework that decouples vision-language understanding from action planning. By integrating a frozen vision-language model, Qwen2.5-VL-7B-Instruct, with lightweight planning logic, we aim to achieve flexible, fast, and adaptable navigation without extensive model fine-tuning. Our framework leverages prompt engineering, structured history management, and a two-frame visual input strategy to enhance decision-making continuity across navigation steps. We evaluate our system on the Room-to-Room benchmark within the VLN-CE setting using the Matterport3D dataset and Habitat-Lab simulation environment. Although our initial results reveal challenges in generalizing to unseen environments under strict evaluation settings, our modular approach lays a foundation for scalable and efficient navigation systems, highlighting promising directions for future improvement through enhanced environmental priors and expanded multimodal input integration.
Authors: Mohammad Jalali, Haoyu Lei, Amin Gohari, Farzan Farnia
Abstract: Diffusion models have demonstrated remarkable success in high-fidelity image synthesis and prompt-guided generative modeling. However, ensuring adequate diversity in generated samples of prompt-guided diffusion models remains a challenge, particularly when the prompts span a broad semantic spectrum and the diversity of generated data needs to be evaluated in a prompt-aware fashion across semantically similar prompts. Recent methods have introduced guidance via diversity measures to encourage more varied generations. In this work, we extend the diversity measure-based approaches by proposing the Scalable Prompt-Aware R\'eny Kernel Entropy Diversity Guidance (SPARKE) method for prompt-aware diversity guidance. SPARKE utilizes conditional entropy for diversity guidance, which dynamically conditions diversity measurement on similar prompts and enables prompt-aware diversity control. While the entropy-based guidance approach enhances prompt-aware diversity, its reliance on the matrix-based entropy scores poses computational challenges in large-scale generation settings. To address this, we focus on the special case of Conditional latent RKE Score Guidance, reducing entropy computation and gradient-based optimization complexity from the $O(n^3)$ of general entropy measures to $O(n)$. The reduced computational complexity allows for diversity-guided sampling over potentially thousands of generation rounds on different prompts. We numerically test the SPARKE method on several text-to-image diffusion models, demonstrating that the proposed method improves the prompt-aware diversity of the generated data without incurring significant computational costs. We release our code on the project page: https://mjalali.github.io/SPARKE
Authors: Mowafaq Salem Alzboon, Mohammad Al-Batah, Muhyeeddin Alqaraleh, Ahmad Abuashour, Ahmad Fuad Bader
Abstract: In many nations, diabetes is becoming a significant health problem, and early identification and control are crucial. Using machine learning algorithms to predict diabetes has yielded encouraging results. Using the Pima Indians Diabetes dataset, this study attempts to evaluate the efficacy of several machine-learning methods for diabetes prediction. The collection includes information on 768 patients, such as their ages, BMIs, and glucose levels. The techniques assessed are Logistic Regression, Decision Tree, Random Forest, k-Nearest Neighbors, Naive Bayes, Support Vector Machine, Gradient Boosting, and Neural Network. The findings indicate that the Neural Network algorithm performed the best, with an accuracy of 78.57 percent, followed by the Random Forest method, with an accuracy of 76.30 percent. The study implies that machine learning algorithms can aid diabetes prediction and be an efficient early detection tool.
Authors: Mohammad Subhi Al-Batah, Mowafaq Salem Alzboon, Muhyeeddin Alqaraleh
Abstract: This study conducts an empirical examination of MLP networks investigated through a rigorous methodical experimentation process involving three diverse datasets: TinyFace, Heart Disease, and Iris. Study Overview: The study includes three key methods: a) a baseline training using the default settings for the Multi-Layer Perceptron (MLP), b) feature selection using Genetic Algorithm (GA) based refinement c) Principal Component Analysis (PCA) based dimension reduction. The results show important information on how such techniques affect performance. While PCA had showed benefits in low-dimensional and noise-free datasets GA consistently increased accuracy in complex datasets by accurately identifying critical features. Comparison reveals that feature selection and dimensionality reduction play interdependent roles in enhancing MLP performance. The study contributes to the literature on feature engineering and neural network parameter optimization, offering practical guidelines for a wide range of machine learning tasks
Authors: Yuhui Ding, Thomas Hofmann
Abstract: Equivariant diffusion models have achieved impressive performance in 3D molecule generation. These models incorporate Euclidean symmetries of 3D molecules by utilizing an SE(3)-equivariant denoising network. However, specialized equivariant architectures limit the scalability and efficiency of diffusion models. In this paper, we propose an approach that relaxes such equivariance constraints. Specifically, our approach learns a sample-dependent SO(3) transformation for each molecule to construct an aligned latent space. A non-equivariant diffusion model is then trained over the aligned representations. Experimental results demonstrate that our approach performs significantly better than previously reported non-equivariant models. It yields sample quality comparable to state-of-the-art equivariant diffusion models and offers improved training and sampling efficiency. Our code is available at https://github.com/skeletondyh/RADM
Authors: Prakamya Mishra, Jiang Liu, Jialian Wu, Xiaodong Yu, Zicheng Liu, Emad Barsoum
Abstract: Large reasoning models (LRMs) have demonstrated impressive reasoning capabilities across a broad range of tasks including Olympiad-level mathematical problems, indicating evidence of their complex reasoning abilities. While many reasoning benchmarks focus on the STEM domain, the ability of LRMs to reason correctly in broader task domains remains underexplored. In this work, we introduce \textbf{TTT-Bench}, a new benchmark that is designed to evaluate basic strategic, spatial, and logical reasoning abilities in LRMs through a suite of four two-player Tic-Tac-Toe-style games that humans can effortlessly solve from a young age. We propose a simple yet scalable programmatic approach for generating verifiable two-player game problems for TTT-Bench. Although these games are trivial for humans, they require reasoning about the intentions of the opponent, as well as the game board's spatial configurations, to ensure a win. We evaluate a diverse set of state-of-the-art LRMs, and \textbf{discover that the models that excel at hard math problems frequently fail at these simple reasoning games}. Further testing reveals that our evaluated reasoning models score on average $\downarrow$ 41\% \& $\downarrow$ 5\% lower on TTT-Bench compared to MATH 500 \& AIME 2024 respectively, with larger models achieving higher performance using shorter reasoning traces, where most of the models struggle on long-term strategic reasoning situations on simple and new TTT-Bench tasks.
Authors: Sajjad Karimi, Amit J. Shah, Gari D. Clifford, Reza Sameni
Abstract: Simultaneous electrocardiography (ECG) and phonocardiogram (PCG) provide a comprehensive, multimodal perspective on cardiac function by capturing the heart's electrical and mechanical activities, respectively. However, the distinct and overlapping information content of these signals, as well as their potential for mutual reconstruction and biomarker extraction, remains incompletely understood, especially under varying physiological conditions and across individuals. In this study, we systematically investigate the common and exclusive characteristics of ECG and PCG using the EPHNOGRAM dataset of simultaneous ECG-PCG recordings during rest and exercise. We employ a suite of linear and nonlinear machine learning models, including non-causal LSTM networks, to reconstruct each modality from the other and analyze the influence of causality, physiological state, and cross-subject variability. Our results demonstrate that nonlinear models, particularly non-causal LSTM, provide superior reconstruction performance, with reconstructing ECG from PCG proving more tractable than the reverse. Exercise and cross-subject scenarios present significant challenges, but envelope-based modeling that utilizes instantaneous amplitude features substantially improves cross-subject generalizability for cross-modal learning. Furthermore, we demonstrate that clinically relevant ECG biomarkers, such as fiducial points and QT intervals, can be estimated from PCG in cross-subject settings. These findings advance our understanding of the relationship between electromechanical cardiac modalities, in terms of both waveform characteristics and the timing of cardiac events, with potential applications in novel multimodal cardiac monitoring technologies.
Authors: Dipanshu Panda, Jayden Koshy Joe, Harshith M R, Swathi Narashiman, Pranay Mathur, Anish Veerakumar, Aniruddh Krishna, Keerthiharan A
Abstract: We present a method for fine-grained control over music generation through inference-time interventions on an autoregressive generative music transformer called MusicGen. Our approach enables timbre transfer, style transfer, and genre fusion by steering the residual stream using weights of linear probes trained on it, or by steering the attention layer activations in a similar manner. We observe that modelling this as a regression task provides improved performance, hypothesizing that the mean-squared-error better preserve meaningful directional information in the activation space. Combined with the global conditioning offered by text prompts in MusicGen, our method provides both global and local control over music generation. Audio samples illustrating our method are available at our demo page.
Authors: Parsa Rahimi, Sebastien Marcel
Abstract: In this paper, we propose ScoreMix, a novel yet simple data augmentation strategy leveraging the score compositional properties of diffusion models to enhance discriminator performance, particularly under scenarios with limited labeled data. By convexly mixing the scores from different class-conditioned trajectories during diffusion sampling, we generate challenging synthetic samples that significantly improve discriminative capabilities in all studied benchmarks. We systematically investigate class-selection strategies for mixing and discover that greater performance gains arise when combining classes distant in the discriminator's embedding space, rather than close in the generator's condition space. Moreover, we empirically show that, under standard metrics, the correlation between the generator's learned condition space and the discriminator's embedding space is minimal. Our approach achieves notable performance improvements without extensive parameter searches, demonstrating practical advantages for training discriminative models while effectively mitigating problems regarding collections of large datasets. Paper website: https://parsa-ra.github.io/scoremix
Authors: Chen-Chia Chang, Wan-Hsuan Lin, Yikang Shen, Yiran Chen, Xin Zhang
Abstract: Automation of analog topology design is crucial due to customized requirements of modern applications with heavily manual engineering efforts. The state-of-the-art work applies a sequence-to-sequence approach and supervised finetuning on language models to generate topologies given user specifications. However, its circuit formulation is inefficient due to O(|V |2) token length and suffers from low precision sensitivity to numeric inputs. In this work, we introduce LaMAGIC2, a succinct float-input canonical formulation with identifier (SFCI) for language model-based analog topology generation. SFCI addresses these challenges by improving component-type recognition through identifier-based representations, reducing token length complexity to O(|V |), and enhancing numeric precision sensitivity for better performance under tight tolerances. Our experiments demonstrate that LaMAGIC2 achieves 34% higher success rates under a tight tolerance of 0.01 and 10X lower MSEs compared to a prior method. LaMAGIC2 also exhibits better transferability for circuits with more vertices with up to 58.5% improvement. These advancements establish LaMAGIC2 as a robust framework for analog topology generation.
Authors: Yeonwoo Jang, Shariqah Hossain, Ashwin Sreevatsa, Diogo Cruz
Abstract: In this work, we show that some machine unlearning methods may fail when subjected to straightforward prompt attacks. We systematically evaluate eight unlearning techniques across three model families, and employ output-based, logit-based, and probe analysis to determine to what extent supposedly unlearned knowledge can be retrieved. While methods like RMU and TAR demonstrate robust unlearning, ELM remains vulnerable to specific prompt attacks (e.g., Hindi filler text in original prompt recovering 57.3% accuracy). Our logit analysis also confirms that unlearned models are generally not hiding knowledge by modifying the way the answer is formatted, as the correlation between output and logit accuracy is strong. These results challenge prevailing assumptions about unlearning effectiveness and highlight the need for evaluation frameworks that can reliably distinguish between true knowledge removal and superficial output suppression. We also publicly make available our evaluation framework to easily evaluate prompting techniques to retrieve unlearning knowledge.
Authors: Iago Alves Brito, Julia Soares Dollis, Fernanda Bufon F\"arber, Diogo Fernandes Costa Silva, Arlindo Rodrigues Galv\~ao Filho
Abstract: We present ToxSyn-PT, the first large-scale Portuguese corpus that enables fine-grained hate-speech classification across nine legally protected minority groups. The dataset contains 53,274 synthetic sentences equally distributed between minorities groups and toxicity labels. ToxSyn-PT is created through a novel four-stage pipeline: (1) a compact, manually curated seed; (2) few-shot expansion with an instruction-tuned LLM; (3) paraphrase-based augmentation; and (4) enrichment, plus additional neutral texts to curb overfitting to group-specific cues. The resulting corpus is class-balanced, stylistically diverse, and free from the social-media domain that dominate existing Portuguese datasets. Despite domain differences with traditional benchmarks, experiments on both binary and multi-label classification on the corpus yields strong results across five public Portuguese hate-speech datasets, demonstrating robust generalization even across domain boundaries. The dataset is publicly released to advance research on synthetic data and hate-speech detection in low-resource settings.
Authors: Andrea Gaggioli, Sabrina Bartolotta, Andrea Ubaldi, Katusha Gerardini, Eleonora Diletta Sarcinella, Alice Chirico
Abstract: Artificial Intelligence holds significant potential to enhance human creativity. However, achieving this vision requires a clearer understanding of how such enhancement can be effectively realized. Adopting the perspective of distributed creativity, we identify three primary modes through which AI can contribute to creative processes: Support, where AI acts as a tool; Synergy, where AI and humans collaborate in complementary ways; and Symbiosis, where human and AI cognition become so integrated that they form a unified creative system. These modes are defined along two key dimensions: the level of technical autonomy exhibited by the AI system and the degree of perceived agency attributed to it. We examine how each configuration influences different levels of creativity - from everyday problem-solving to paradigm-shifting innovation - and discuss the theoretical, ethical, and design implications.
Authors: Andrea Yaoyun Cui, Pengfei Yu
Abstract: Language models are essentially probability distributions over token sequences. Auto-regressive models generate sentences by iteratively computing and sampling from the distribution of the next token. This iterative sampling introduces stochasticity, leading to the assumption that language models make probabilistic decisions, similar to sampling from unknown distributions. Building on this assumption, prior research has used simulated Gibbs sampling, inspired by experiments designed to elicit human priors, to infer the priors of language models. In this paper, we revisit a critical question: Do language models possess Bayesian brains? Our findings show that under certain conditions, language models can exhibit near-deterministic decision-making, such as producing maximum likelihood estimations, even with a non-zero sampling temperature. This challenges the sampling assumption and undermines previous methods for eliciting human-like priors. Furthermore, we demonstrate that without proper scrutiny, a system with deterministic behavior undergoing simulated Gibbs sampling can converge to a "false prior." To address this, we propose a straightforward approach to distinguish between stochastic and deterministic decision patterns in Gibbs sampling, helping to prevent the inference of misleading language model priors. We experiment on a variety of large language models to identify their decision patterns under various circumstances. Our results provide key insights in understanding decision making of large language models.
Authors: Pooneh Mousavi, Gallil Maimon, Adel Moumen, Darius Petermann, Jiatong Shi, Haibin Wu, Haici Yang, Anastasia Kuznetsova, Artem Ploujnikov, Ricard Marxer, Bhuvana Ramabhadran, Benjamin Elizalde, Loren Lugosch, Jinyu Li, Cem Subakan, Phil Woodland, Minje Kim, Hung-yi Lee, Shinji Watanabe, Yossi Adi, Mirco Ravanelli
Abstract: Discrete audio tokens are compact representations that aim to preserve perceptual quality, phonetic content, and speaker characteristics while enabling efficient storage and inference, as well as competitive performance across diverse downstream tasks.They provide a practical alternative to continuous features, enabling the integration of speech and audio into modern large language models (LLMs). As interest in token-based audio processing grows, various tokenization methods have emerged, and several surveys have reviewed the latest progress in the field. However, existing studies often focus on specific domains or tasks and lack a unified comparison across various benchmarks. This paper presents a systematic review and benchmark of discrete audio tokenizers, covering three domains: speech, music, and general audio. We propose a taxonomy of tokenization approaches based on encoder-decoder, quantization techniques, training paradigm, streamability, and application domains. We evaluate tokenizers on multiple benchmarks for reconstruction, downstream performance, and acoustic language modeling, and analyze trade-offs through controlled ablation studies. Our findings highlight key limitations, practical considerations, and open challenges, providing insight and guidance for future research in this rapidly evolving area. For more information, including our main results and tokenizer database, please refer to our website: https://poonehmousavi.github.io/dates-website/.
Authors: Zige Wang, Qi Zhu, Fei Mi, Minghui Xu, Ruochun Jin, Wenjing Yang
Abstract: Gradient-based data influence approximation has been leveraged to select useful data samples in the supervised fine-tuning of large language models. However, the computation of gradients throughout the fine-tuning process requires too many resources to be feasible in practice. In this paper, we propose an efficient gradient-based data selection framework with clustering and a modified Upper Confidence Bound (UCB) algorithm. Based on the intuition that data samples with similar gradient features will have similar influences, we first perform clustering on the training data pool. Then, we frame the inter-cluster data selection as a constrained computing budget allocation problem and consider it a multi-armed bandit problem. A modified UCB algorithm is leveraged to solve this problem. Specifically, during the iterative sampling process, historical data influence information is recorded to directly estimate the distributions of each cluster, and a cold start is adopted to balance exploration and exploitation. Experimental results on various benchmarks show that our proposed framework, ClusterUCB, can achieve comparable results to the original gradient-based data selection methods while greatly reducing computing consumption.
Authors: Yisi Liu, Chenyang Wang, Hanjo Kim, Raniya Khan, Gopala Anumanchipalli
Abstract: Voice conversion has emerged as a pivotal technology in numerous applications ranging from assistive communication to entertainment. In this paper, we present RT-VC, a zero-shot real-time voice conversion system that delivers ultra-low latency and high-quality performance. Our approach leverages an articulatory feature space to naturally disentangle content and speaker characteristics, facilitating more robust and interpretable voice transformations. Additionally, the integration of differentiable digital signal processing (DDSP) enables efficient vocoding directly from articulatory features, significantly reducing conversion latency. Experimental evaluations demonstrate that, while maintaining synthesis quality comparable to the current state-of-the-art (SOTA) method, RT-VC achieves a CPU latency of 61.4 ms, representing a 13.3\% reduction in latency.
Authors: Ali Almutairi, Abdullah Alsuhaibani, Shoaib Jameel, Usman Naseem, Gelareh Mohammadi, Imran Razzak
Abstract: Training deep learning networks with minimal supervision has gained significant research attention due to its potential to reduce reliance on extensive labelled data. While self-training methods have proven effective in semi-supervised learning, they remain vulnerable to errors from noisy pseudo labels. Moreover, most recent approaches to the few-label classification problem are either designed for resource-rich languages such as English or involve complex cascading models that are prone to overfitting. To address the persistent challenge of few-label text classification in truly low-resource linguistic contexts, where existing methods often struggle with noisy pseudo-labels and domain adaptation, we propose Flick. Unlike prior methods that rely on generic multi-cluster pseudo-labelling or complex cascading architectures, Flick leverages the fundamental insight that distilling high-confidence pseudo-labels from a broader set of initial clusters can dramatically improve pseudo-label quality, particularly for linguistically diverse, low-resource settings. Flick introduces a novel pseudo-label refinement component, a departure from traditional pseudo-labelling strategies by identifying and leveraging top-performing pseudo-label clusters. This component specifically learns to distil highly reliable pseudo-labels from an initial broad set by focusing on single-cluster cohesion and leveraging an adaptive top-k selection mechanism. This targeted refinement process is crucial for mitigating the propagation of errors inherent in low-resource data, allowing for robust fine-tuning of pre-trained language models with only a handful of true labels. We demonstrate Flick's efficacy across 14 diverse datasets, encompassing challenging low-resource languages such as Arabic, Urdu, and Setswana, alongside English, showcasing its superior performance and adaptability.
Authors: Hossein A. Rahmani, Varsha Ramineni, Nick Craswell, Bhaskar Mitra, Emine Yilmaz
Abstract: Test collections are crucial for evaluating Information Retrieval (IR) systems. Creating a diverse set of user queries for these collections can be challenging, and obtaining relevance judgments, which indicate how well retrieved documents match a query, is often costly and resource-intensive. Recently, generating synthetic datasets using Large Language Models (LLMs) has gained attention in various applications. While previous work has used LLMs to generate synthetic queries or documents to improve ranking models, using LLMs to create synthetic test collections is still relatively unexplored. Previous work~\cite{rahmani2024synthetic} showed that synthetic test collections have the potential to be used for system evaluation, however, more analysis is needed to validate this claim. In this paper, we thoroughly investigate the reliability of synthetic test collections constructed using LLMs, where LLMs are used to generate synthetic queries, labels, or both. In particular, we examine the potential biases that might occur when such test collections are used for evaluation. We first empirically show the presence of such bias in evaluation results and analyse the effects it might have on system evaluation. We further validate the presence of such bias using a linear mixed-effects model. Our analysis shows that while the effect of bias present in evaluation results obtained using synthetic test collections could be significant, for e.g.~computing absolute system performance, its effect may not be as significant in comparing relative system performance. Codes and data are available at: https://github.com/rahmanidashti/BiasSyntheticData.
Authors: Hamzeh Asgharnezhad, Pegah Tabarisaadi, Abbas Khosravi, Roohallah Alizadehsani, U. Rajendra Acharya
Abstract: Accurate and reliable skin cancer diagnosis is critical for early treatment and improved patient outcomes. Deep learning (DL) models have shown promise in automating skin cancer classification, but their performance can be limited by data scarcity and a lack of uncertainty awareness. In this study, we present a comprehensive evaluation of DL-based skin lesion classification using transfer learning and uncertainty quantification (UQ) on the HAM10000 dataset. In the first phase, we benchmarked several pre-trained feature extractors-including Contrastive Language-Image Pretraining (CLIP) variants, Residual Network-50 (ResNet50), Densely Connected Convolutional Network (DenseNet121), Visual Geometry Group network (VGG16), and EfficientNet-V2-Large-combined with a range of traditional classifiers such as Support Vector Machine (SVM), eXtreme Gradient Boosting (XGBoost), and logistic regression. Our results show that CLIP-based vision transformers, particularly LAION CLIP ViT-H/14 with SVM, deliver the highest classification performance. In the second phase, we incorporated UQ using Monte Carlo Dropout (MCD), Ensemble, and Ensemble Monte Carlo Dropout (EMCD) to assess not only prediction accuracy but also the reliability of model outputs. We evaluated these models using uncertainty-aware metrics such as uncertainty accuracy(UAcc), uncertainty sensitivity(USen), uncertainty specificity(USpe), and uncertainty precision(UPre). The results demonstrate that ensemble methods offer a good trade-off between accuracy and uncertainty handling, while EMCD is more sensitive to uncertain predictions. This study highlights the importance of integrating UQ into DL-based medical diagnosis to enhance both performance and trustworthiness in real-world clinical applications.
Authors: Yuliang Zhu, Jing Cheng, Qi Xie, Zhuo-Xu Cui, Qingyong Zhu, Yuanyuan Liu, Xin Liu, Jianfeng Ren, Chengbo Wang, Dong Liang
Abstract: Dynamic Magnetic Resonance Imaging (MRI) exhibits transformation symmetries, including spatial rotation symmetry within individual frames and temporal symmetry along the time dimension. Explicit incorporation of these symmetry priors in the reconstruction model can significantly improve image quality, especially under aggressive undersampling scenarios. Recently, Equivariant convolutional neural network (ECNN) has shown great promise in exploiting spatial symmetry priors. However, existing ECNNs critically fail to model temporal symmetry, arguably the most universal and informative structural prior in dynamic MRI reconstruction. To tackle this issue, we propose a novel Deep Unrolling Network with Spatiotemporal Rotation Equivariance (DUN-SRE) for Dynamic MRI Reconstruction. The DUN-SRE establishes spatiotemporal equivariance through a (2+1)D equivariant convolutional architecture. In particular, it integrates both the data consistency and proximal mapping module into a unified deep unrolling framework. This architecture ensures rigorous propagation of spatiotemporal rotation symmetry constraints throughout the reconstruction process, enabling more physically accurate modeling of cardiac motion dynamics in cine MRI. In addition, a high-fidelity group filter parameterization mechanism is developed to maintain representation precision while enforcing symmetry constraints. Comprehensive experiments on Cardiac CINE MRI datasets demonstrate that DUN-SRE achieves state-of-the-art performance, particularly in preserving rotation-symmetric structures, offering strong generalization capability to a broad range of dynamic MRI reconstruction tasks.
Authors: Akshar Tumu, Henrik I. Christensen, Marcell Vazquez-Chanlatte, Chikao Tsuchiya, Dhaval Bhanderi
Abstract: Lane-topology prediction is a critical component of safe and reliable autonomous navigation. An accurate understanding of the road environment aids this task. We observe that this information often follows conventions encoded in natural language, through design codes that reflect the road structure and road names that capture the road functionality. We augment this information in a lightweight manner to SMERF, a map-prior-based online lane-topology prediction model, by combining structured road metadata from OSM maps and lane-width priors from Road design manuals with the road centerline encodings. We evaluate our method on two geo-diverse complex intersection scenarios. Our method shows improvement in both lane and traffic element detection and their association. We report results using four topology-aware metrics to comprehensively assess the model performance. These results demonstrate the ability of our approach to generalize and scale to diverse topologies and conditions.
Authors: Sadia Kamal, Tim Oates, Joy Wan
Abstract: Skin carcinoma is the most prevalent form of cancer globally, accounting for over $8 billion in annual healthcare expenditures. In clinical settings, physicians document patient visits using detailed SOAP (Subjective, Objective, Assessment, and Plan) notes. However, manually generating these notes is labor-intensive and contributes to clinician burnout. In this work, we propose a weakly supervised multimodal framework to generate clinically structured SOAP notes from limited inputs, including lesion images and sparse clinical text. Our approach reduces reliance on manual annotations, enabling scalable, clinically grounded documentation while alleviating clinician burden and reducing the need for large annotated data. Our method achieves performance comparable to GPT-4o, Claude, and DeepSeek Janus Pro across key clinical relevance metrics. To evaluate clinical quality, we introduce two novel metrics MedConceptEval and Clinical Coherence Score (CCS) which assess semantic alignment with expert medical concepts and input features, respectively.
Authors: Seyed Moein Abtahi, Akramul Azim
Abstract: This study examined code issue detection and revision automation by integrating Large Language Models (LLMs) such as OpenAI's GPT-3.5 Turbo and GPT-4o into software development workflows. A static code analysis framework detects issues such as bugs, vulnerabilities, and code smells within a large-scale software project. Detailed information on each issue was extracted and organized to facilitate automated code revision using LLMs. An iterative prompt engineering process is applied to ensure that prompts are structured to produce accurate and organized outputs aligned with the project requirements. Retrieval-augmented generation (RAG) is implemented to enhance the relevance and precision of the revisions, enabling LLM to access and integrate real-time external knowledge. The issue of LLM hallucinations - where the model generates plausible but incorrect outputs - is addressed by a custom-built "Code Comparison App," which identifies and corrects erroneous changes before applying them to the codebase. Subsequent scans using the static code analysis framework revealed a significant reduction in code issues, demonstrating the effectiveness of combining LLMs, static analysis, and RAG to improve code quality, streamline the software development process, and reduce time and resource expenditure.
Authors: Deliang Wang, Chao Yang, Gaowei Chen
Abstract: Students' academic emotions significantly influence their social behavior and learning performance. Traditional approaches to automatically and accurately analyze these emotions have predominantly relied on supervised machine learning algorithms. However, these models often struggle to generalize across different contexts, necessitating repeated cycles of data collection, annotation, and training. The emergence of Vision-Language Models (VLMs) offers a promising alternative, enabling generalization across visual recognition tasks through zero-shot prompting without requiring fine-tuning. This study investigates the potential of VLMs to analyze students' academic emotions via facial expressions in an online learning environment. We employed two VLMs, Llama-3.2-11B-Vision-Instruct and Qwen2.5-VL-7B-Instruct, to analyze 5,000 images depicting confused, distracted, happy, neutral, and tired expressions using zero-shot prompting. Preliminary results indicate that both models demonstrate moderate performance in academic facial expression recognition, with Qwen2.5-VL-7B-Instruct outperforming Llama-3.2-11B-Vision-Instruct. Notably, both models excel in identifying students' happy emotions but fail to detect distracted behavior. Additionally, Qwen2.5-VL-7B-Instruct exhibits relatively high performance in recognizing students' confused expressions, highlighting its potential for practical applications in identifying content that causes student confusion.
Authors: Jun Yin, Jing Zhong, Peilin Li, Pengyu Zeng, Miao Zhang, Ran Luo, Shuai Lu
Abstract: Urban cultures and architectural styles vary significantly across cities due to geographical, chronological, historical, and socio-political factors. Understanding these differences is essential for anticipating how cities may evolve in the future. As representative cases of historical continuity and modern innovation in China, Beijing and Shenzhen offer valuable perspectives for exploring the transformation of urban streetscapes. However, conventional approaches to urban cultural studies often rely on expert interpretation and historical documentation, which are difficult to standardize across different contexts. To address this, we propose a multimodal research framework based on vision-language models, enabling automated and scalable analysis of urban streetscape style differences. This approach enhances the objectivity and data-driven nature of urban form research. The contributions of this study are as follows: First, we construct UrbanDiffBench, a curated dataset of urban streetscapes containing architectural images from different periods and regions. Second, we develop UrbanSense, the first vision-language-model-based framework for urban streetscape analysis, enabling the quantitative generation and comparison of urban style representations. Third, experimental results show that Over 80% of generated descriptions pass the t-test (p less than 0.05). High Phi scores (0.912 for cities, 0.833 for periods) from subjective evaluations confirm the method's ability to capture subtle stylistic differences. These results highlight the method's potential to quantify and interpret urban style evolution, offering a scientifically grounded lens for future design.
Authors: Dongwon Jung, Wenxuan Zhou, Muhao Chen
Abstract: Training large language models (LLMs) with chain-of-thought (CoT) supervision has proven effective for enhancing their reasoning abilities. However, obtaining reliable and accurate reasoning supervision remains a significant challenge. We propose a scalable method for generating a high-quality CoT supervision dataset by leveraging the determinism of program execution. Unlike existing reasoning dataset generation methods that rely on costly human annotations or error-prone LLM-generated CoT, our approach extracts verifiable, step-by-step reasoning traces from code execution and transforms them into a natural language CoT reasoning. Experiments on reasoning benchmarks across various domains show that our method effectively equips LLMs with transferable reasoning abilities across diverse tasks. Furthermore, the ablation studies validate that our method produces highly accurate reasoning data and reduces overall token length during inference by reducing meaningless repetition and overthinking.
Authors: Yanlong Chen, Mattia Orlandi, Pierangelo Maria Rapa, Simone Benatti, Luca Benini, Yawei Li
Abstract: Physiological signals are often corrupted by motion artifacts, baseline drift, and other low-SNR disturbances, which pose significant challenges for analysis. Additionally, these signals exhibit strong non-stationarity, with sharp peaks and abrupt changes that evolve continuously, making them difficult to represent using traditional time-domain or filtering methods. To address these issues, a novel wavelet-based approach for physiological signal analysis is presented, aiming to capture multi-scale time-frequency features in various physiological signals. Leveraging this technique, two large-scale pretrained models specific to EMG and ECG are introduced for the first time, achieving superior performance and setting new baselines in downstream tasks. Additionally, a unified multi-modal framework is constructed by integrating pretrained EEG model, where each modality is guided through its dedicated branch and fused via learnable weighted fusion. This design effectively addresses challenges such as low signal-to-noise ratio, high inter-subject variability, and device mismatch, outperforming existing methods on multi-modal tasks. The proposed wavelet-based architecture lays a solid foundation for analysis of diverse physiological signals, while the multi-modal design points to next-generation physiological signal processing with potential impact on wearable health monitoring, clinical diagnostics, and broader biomedical applications.
Authors: Jikai Jin, Vasilis Syrgkanis, Sham Kakade, Hanlin Zhang
Abstract: Faithful evaluation of language model capabilities is crucial for deriving actionable insights that can inform model development. However, rigorous causal evaluations in this domain face significant methodological challenges, including complex confounding effects and prohibitive computational costs associated with extensive retraining. To tackle these challenges, we propose a causal representation learning framework wherein observed benchmark performance is modeled as a linear transformation of a few latent capability factors. Crucially, these latent factors are identified as causally interrelated after appropriately controlling for the base model as a common confounder. Applying this approach to a comprehensive dataset encompassing over 1500 models evaluated across six benchmarks from the Open LLM Leaderboard, we identify a concise three-node linear causal structure that reliably explains the observed performance variations. Further interpretation of this causal structure provides substantial scientific insights beyond simple numerical rankings: specifically, we reveal a clear causal direction starting from general problem-solving capabilities, advancing through instruction-following proficiency, and culminating in mathematical reasoning ability. Our results underscore the essential role of carefully controlling base model variations during evaluation, a step critical to accurately uncovering the underlying causal relationships among latent model capabilities.
Authors: Zhiyang Xu, Jiuhai Chen, Zhaojiang Lin, Xichen Pan, Lifu Huang, Tianyi Zhou, Madian Khabsa, Qifan Wang, Di Jin, Michihiro Yasunaga, Lili Yu, Xi Victoria Lin, Shaoliang Nie
Abstract: Recent advances in large language models (LLMs) have enabled multimodal foundation models to tackle both image understanding and generation within a unified framework. Despite these gains, unified models often underperform compared to specialized models in either task. A key challenge in developing unified models lies in the inherent differences between the visual features needed for image understanding versus generation, as well as the distinct training processes required for each modality. In this work, we introduce Pisces, an auto-regressive multimodal foundation model that addresses this challenge through a novel decoupled visual encoding architecture and tailored training techniques optimized for multimodal generation. Combined with meticulous data curation, pretraining, and finetuning, Pisces achieves competitive performance in both image understanding and image generation. We evaluate Pisces on over 20 public benchmarks for image understanding, where it demonstrates strong performance across a wide range of tasks. Additionally, on GenEval, a widely adopted benchmark for image generation, Pisces exhibits robust generative capabilities. Our extensive analysis reveals the synergistic relationship between image understanding and generation, and the benefits of using separate visual encoders, advancing the field of unified multimodal models.
Authors: Jiaqi Lv, Xufeng He, Yanchen Liu, Xu Dai, Yang Hu, Shouyi Yin
Abstract: The rapid growth of deep learning has driven exponential increases in model parameters and computational demands. NVIDIA GPUs and their CUDA-based software ecosystem provide robust support for parallel computing, significantly alleviating computational bottlenecks. Meanwhile, due to the cultivation of user programming habits and the high performance of GPUs, the CUDA ecosystem has established a dominant position in the field of parallel software. This dominance requires other hardware platforms to support CUDA-based software with performance portability. However, translating CUDA code to other platforms poses significant challenges due to differences in parallel programming paradigms and hardware architectures. Existing approaches rely on language extensions, domain-specific languages (DSLs), or compilers but face limitations in workload coverage and generalizability. Moreover, these methods often incur substantial development costs. Recently, LLMs have demonstrated extraordinary potential in various vertical domains, especially in code-related tasks. However, the performance of existing LLMs in CUDA transpilation, particularly for high-performance code, remains suboptimal. The main reason for this limitation lies in the lack of high-quality training datasets. To address these challenges, we propose a novel framework for generating high-performance CUDA and corresponding platform code pairs, leveraging AI compiler and automatic optimization technology. We further enhance the framework with a graph-based data augmentation method and introduce HPCTransEval, a benchmark for evaluating LLM performance on CUDA transpilation. We conduct experiments using CUDA-to-CPU transpilation as a case study on leading LLMs. The result demonstrates that our framework significantly improves CUDA transpilation, highlighting the potential of LLMs to address compatibility challenges within the CUDA ecosystem.
Authors: Tzu-Heng Huang, Harit Vishwakarma, Frederic Sala
Abstract: Large language models (LLMs) are widely used to evaluate the quality of LLM generations and responses, but this leads to significant challenges: high API costs, uncertain reliability, inflexible pipelines, and inherent biases. To address these, we introduce PAJAMA (Program-As-a-Judge for Automated Model Assessment), a new alternative that uses LLMs to synthesize executable judging programs instead of directly scoring responses. These synthesized programs can be stored and run locally, costing orders of magnitude less while providing interpretable, and auditable judging logic that can be easily adapted. Program-based judges mitigate biases, improving judgment consistency by 15.83% and reducing biased responses by 23.7% on average compared to a Qwen2.5-14B-based LLM-as-a-judge. When program judgments are distilled into a model, PAJAMA outperforms LLM-as-a-judge on the challenging CHAT-HARD subset of RewardBench, outperforming metrics by 2.19% on Prometheus and 8.67% on the JudgeLM dataset, all at three orders of magnitude lower cost.
Authors: Yuhua Jiang, Yuwen Xiong, Yufeng Yuan, Chao Xin, Wenyuan Xu, Yu Yue, Qianchuan Zhao, Lin Yan
Abstract: Large Language Models (LLMs) have demonstrated impressive capabilities in complex reasoning tasks, yet they still struggle to reliably verify the correctness of their own outputs. Existing solutions to this verification challenge often depend on separate verifier models or require multi-stage self-correction training pipelines, which limit scalability. In this paper, we propose Policy as Generative Verifier (PAG), a simple and effective framework that empowers LLMs to self-correct by alternating between policy and verifier roles within a unified multi-turn reinforcement learning (RL) paradigm. Distinct from prior approaches that always generate a second attempt regardless of model confidence, PAG introduces a selective revision mechanism: the model revises its answer only when its own generative verification step detects an error. This verify-then-revise workflow not only alleviates model collapse but also jointly enhances both reasoning and verification abilities. Extensive experiments across diverse reasoning benchmarks highlight PAG's dual advancements: as a policy, it enhances direct generation and self-correction accuracy; as a verifier, its self-verification outperforms self-consistency.
Authors: Daizhan Cheng
Abstract: The semi-tensor product (STP) of vectors is a generalization of conventional inner product of vectors, which allows the factor vectors to of different dimensions. This paper proposes a domain-based convolutional product (CP). Combining domain-based CP with STP of vectors, a new CP is proposed. Since there is no zero or any other padding, it can avoid the junk information caused by padding. Using it, the STP-based convolutional neural network (CNN) is developed. Its application to image and third order signal identifications is considered.
Authors: Ching Chang, Jeehyun Hwang, Yidan Shi, Haixin Wang, Wen-Chih Peng, Tien-Fu Chen, Wei Wang
Abstract: Time series data in real-world applications such as healthcare, climate modeling, and finance are often irregular, multimodal, and messy, with varying sampling rates, asynchronous modalities, and pervasive missingness. However, existing benchmarks typically assume clean, regularly sampled, unimodal data, creating a significant gap between research and real-world deployment. We introduce Time-IMM, a dataset specifically designed to capture cause-driven irregularity in multimodal multivariate time series. Time-IMM represents nine distinct types of time series irregularity, categorized into trigger-based, constraint-based, and artifact-based mechanisms. Complementing the dataset, we introduce IMM-TSF, a benchmark library for forecasting on irregular multimodal time series, enabling asynchronous integration and realistic evaluation. IMM-TSF includes specialized fusion modules, including a timestamp-to-text fusion module and a multimodality fusion module, which support both recency-aware averaging and attention-based integration strategies. Empirical results demonstrate that explicitly modeling multimodality on irregular time series data leads to substantial gains in forecasting performance. Time-IMM and IMM-TSF provide a foundation for advancing time series analysis under real-world conditions. The dataset is publicly available at https://www.kaggle.com/datasets/blacksnail789521/time-imm/data, and the benchmark library can be accessed at https://anonymous.4open.science/r/IMMTSF_NeurIPS2025.
URLs: https://www.kaggle.com/datasets/blacksnail789521/time-imm/data,, https://anonymous.4open.science/r/IMMTSF_NeurIPS2025.
Authors: Tony Alex, Wish Suharitdamrong, Sara Atito, Armin Mustafa, Philip J. B. Jackson, Imran Razzak, Muhammad Awais
Abstract: The integration of audio perception capabilities into Large Language Models (LLMs) has enabled significant advances in Audio-LLMs. Although application-focused developments, particularly in curating training data for specific capabilities e.g., audio reasoning, have progressed rapidly, the underlying mechanisms that govern efficient transfer of rich semantic representations from audio encoders to LLMs remain under-explored. We conceptualize effective audio-LLM interaction as the LLM's ability to proficiently probe the audio encoder representations to satisfy textual queries. This paper presents a systematic investigation on how architectural design choices can affect that. Beginning with a standard Pengi/LLaVA-style audio-LLM architecture, we propose and evaluate several modifications guided by hypotheses derived from mechanistic interpretability studies and LLM operational principles. Our experiments demonstrate that: (1) delaying audio integration until the LLM's initial layers establish textual context that enhances its ability to probe the audio representations for relevant information; (2) the LLM can proficiently probe audio representations exclusively through LLM layer's attention submodule, without requiring propagation to its Feed-Forward Network (FFN) submodule; (3) an efficiently integrated ensemble of diverse audio encoders provides richer, complementary representations, thereby broadening the LLM's capacity to probe a wider spectrum of audio information. All hypotheses are evaluated using an identical three-stage training curriculum on a dataset of 5.6 million audio-text pairs, ensuring controlled comparisons. Our final architecture, which incorporates all proposed modifications, achieves relative improvements from 10\% to 60\% over the baseline, validating our approach to optimizing cross-modal information transfer in audio-LLMs. Project page: https://ta012.github.io/PAL/
Authors: Kaiyuan Zhang, Siyuan Cheng, Hanxi Guo, Yuetian Chen, Zian Su, Shengwei An, Yuntao Du, Charles Fleming, Ashish Kundu, Xiangyu Zhang, Ninghui Li
Abstract: Large language models (LLMs) have achieved remarkable success and are widely adopted for diverse applications. However, fine-tuning these models often involves private or sensitive information, raising critical privacy concerns. In this work, we conduct the first comprehensive study evaluating the vulnerability of fine-tuned LLMs to membership inference attacks (MIAs). Our empirical analysis demonstrates that MIAs exploit the loss reduction during fine-tuning, making them highly effective in revealing membership information. These findings motivate the development of our defense. We propose SOFT (\textbf{S}elective data \textbf{O}bfuscation in LLM \textbf{F}ine-\textbf{T}uning), a novel defense technique that mitigates privacy leakage by leveraging influential data selection with an adjustable parameter to balance utility preservation and privacy protection. Our extensive experiments span six diverse domains and multiple LLM architectures and scales. Results show that SOFT effectively reduces privacy risks while maintaining competitive model performance, offering a practical and scalable solution to safeguard sensitive information in fine-tuned LLMs.
Authors: Toby Walsh
Abstract: We consider strategy proof mechanisms for facility location which maximize equitability between agents. As is common in the literature, we measure equitability with the Gini index. We first prove a simple but fundamental impossibility result that no strategy proof mechanism can bound the approximation ratio of the optimal Gini index of utilities for one or more facilities. We propose instead computing approximation ratios of the complemented Gini index of utilities, and consider how well both deterministic and randomized mechanisms approximate this. In addition, as Nash welfare is often put forwards as an equitable compromise between egalitarian and utilitarian outcomes, we consider how well mechanisms approximate the Nash welfare.
Authors: Stone Yun, Alexander Wong
Abstract: Deep neural network (DNN) quantization for fast, efficient inference has been an important tool in limiting the cost of machine learning (ML) model inference. Quantization-specific model development techniques such as regularization, quantization-aware training, and quantization-robustness penalties have served to greatly boost the accuracy and robustness of modern DNNs. However, very little exploration has been done on improving the initial conditions of DNN training for quantization. Just as random weight initialization has been shown to significantly impact test accuracy of floating point models, it would make sense that different weight initialization methods impact quantization robustness of trained models. We present an extensive study examining the effects of different weight initializations on a variety of CNN building blocks commonly used in efficient CNNs. This analysis reveals that even with varying CNN architectures, the choice of random weight initializer can significantly affect final quantization robustness. Next, we explore a new method for quantization-robust CNN initialization -- using Graph Hypernetworks (GHN) to predict parameters of quantized DNNs. Besides showing that GHN-predicted parameters are quantization-robust after regular float32 pretraining (of the GHN), we find that finetuning GHNs to predict parameters for quantized graphs (which we call GHN-QAT) can further improve quantized accuracy of CNNs. Notably, GHN-QAT shows significant accuracy improvements for even 4-bit quantization and better-than-random accuracy for 2-bits. To the best of our knowledge, this is the first in-depth study on quantization-aware DNN weight initialization. GHN-QAT offers a novel approach to quantized DNN model design. Future investigations, such as using GHN-QAT-initialized parameters for quantization-aware training, can further streamline the DNN quantization process.
Authors: Felix H\"arer
Abstract: Recent advancements in LLMs indicate potential for novel applications, e.g., through reasoning capabilities in the latest OpenAI and DeepSeek models. For applying these models in specific domains beyond text generation, LLM-based multi-agent approaches can be utilized that solve complex tasks by combining reasoning techniques, code generation, and software execution. Applications might utilize these capabilities and the knowledge of specialized LLM agents. However, while many evaluations are performed on LLMs, reasoning techniques, and applications individually, their joint specification and combined application is not explored well. Defined specifications for multi-agent LLM systems are required to explore their potential and their suitability for specific applications, allowing for systematic evaluations of LLMs, reasoning techniques, and related aspects. This paper reports the results of exploratory research to specify and evaluate these aspects through a multi-agent system. The system architecture and prototype are extended from previous research and a specification is introduced for multi-agent systems. Test cases involving cybersecurity tasks indicate feasibility of the architecture and evaluation approach. In particular, the results show the evaluation of question answering, server security, and network security tasks that were completed correctly by agents with LLMs from OpenAI and DeepSeek.
Authors: Shuyang Li, Shuang Wang, Zhuangzhuang Sun, Jing Xiao
Abstract: The Reference Remote Sensing Image Segmentation (RRSIS) task generates segmentation masks for specified objects in images based on textual descriptions, which has attracted widespread attention and research interest. Current RRSIS methods rely on multi-modal fusion backbones and semantic segmentation heads but face challenges like dense annotation requirements and complex scene interpretation. To address these issues, we propose a framework named \textit{prompt-generated semantic localization guiding Segment Anything Model}(PSLG-SAM), which decomposes the RRSIS task into two stages: coarse localization and fine segmentation. In coarse localization stage, a visual grounding network roughly locates the text-described object. In fine segmentation stage, the coordinates from the first stage guide the Segment Anything Model (SAM), enhanced by a clustering-based foreground point generator and a mask boundary iterative optimization strategy for precise segmentation. Notably, the second stage can be train-free, significantly reducing the annotation data burden for the RRSIS task. Additionally, decomposing the RRSIS task into two stages allows for focusing on specific region segmentation, avoiding interference from complex scenes.We further contribute a high-quality, multi-category manually annotated dataset. Experimental validation on two datasets (RRSIS-D and RRSIS-M) demonstrates that PSLG-SAM achieves significant performance improvements and surpasses existing state-of-the-art models.Our code will be made publicly available.
Authors: Sangmin Song, Juhwan Choi, JungMin Yun, YoungBin Kim
Abstract: Large language models (LLMs) have demonstrated remarkable performance in zero-shot dialogue state tracking (DST), reducing the need for task-specific training. However, conventional DST benchmarks primarily focus on structured user-agent conversations, failing to capture the complexities of real-world multi-user interactions. In this study, we assess the robustness of LLMs in multi-user DST while minimizing dataset construction costs. Inspired by recent advances in LLM-based data annotation, we extend an existing DST dataset by generating utterances of a second user based on speech act theory. Our methodology systematically incorporates a second user's utterances into conversations, enabling a controlled evaluation of LLMs in multi-user settings. Experimental results reveal a significant performance drop compared to single-user DST, highlighting the limitations of current LLMs in extracting and tracking dialogue states amidst multiple speakers. Our findings emphasize the need for future research to enhance LLMs for multi-user DST scenarios, paving the way for more realistic and robust DST models.
Authors: Yilin Xiao, Chuang Zhou, Qinggang Zhang, Bo Li, Qing Li, Xiao Huang
Abstract: Large language models (LLMs) often struggle with knowledge-intensive tasks due to a lack of background knowledge and a tendency to hallucinate. To address these limitations, integrating knowledge graphs (KGs) with LLMs has been intensively studied. Existing KG-enhanced LLMs focus on supplementary factual knowledge, but still struggle with solving complex questions. We argue that refining the relationships among facts and organizing them into a logically consistent reasoning path is equally important as factual knowledge itself. Despite their potential, extracting reliable reasoning paths from KGs poses the following challenges: the complexity of graph structures and the existence of multiple generated paths, making it difficult to distinguish between useful and redundant ones. To tackle these challenges, we propose the RRP framework to mine the knowledge graph, which combines the semantic strengths of LLMs with structural information obtained through relation embedding and bidirectional distribution learning. Additionally, we introduce a rethinking module that evaluates and refines reasoning paths according to their significance. Experimental results on two public datasets show that RRP achieves state-of-the-art performance compared to existing baseline methods. Moreover, RRP can be easily integrated into various LLMs to enhance their reasoning abilities in a plug-and-play manner. By generating high-quality reasoning paths tailored to specific questions, RRP distills effective guidance for LLM reasoning.
Authors: Zicheng Zhao, Kangyu Wang, Shijie Li, Rui Qian, Weiyao Lin, Huabin Liu
Abstract: Despite advancements in Video Large Language Models (Vid-LLMs) improving multimodal understanding, challenges persist in streaming video reasoning due to its reliance on contextual information. Existing paradigms feed all available historical contextual information into Vid-LLMs, resulting in a significant computational burden for visual data processing. Furthermore, the inclusion of irrelevant context distracts models from key details. This paper introduces a challenging task called Context-guided Streaming Video Reasoning (CogStream), which simulates real-world streaming video scenarios, requiring models to identify the most relevant historical contextual information to deduce answers for questions about the current stream. To support CogStream, we present a densely annotated dataset featuring extensive and hierarchical question-answer pairs, generated by a semi-automatic pipeline. Additionally, we present CogReasoner as a baseline model. It efficiently tackles this task by leveraging visual stream compression and historical dialogue retrieval. Extensive experiments prove the effectiveness of this method. Code will be released soon.
Authors: Xiaolin Hu, Qinghua Zhou, Bogdan Grechuk, Ivan Y. Tyukin
Abstract: Interactive theorem provers (ITPs) are powerful tools for the formal verification of mathematical proofs down to the axiom level. However, their lack of a natural language interface remains a significant limitation. Recent advancements in large language models (LLMs) have enhanced the understanding of natural language inputs, paving the way for autoformalization - the process of translating natural language proofs into formal proofs that can be verified. Despite these advancements, existing autoformalization approaches are limited to verifying complete proofs and lack the capability for finer, sentence-level verification. To address this gap, we propose StepProof, a novel autoformalization method designed for granular, step-by-step verification. StepProof breaks down complete proofs into multiple verifiable subproofs, enabling sentence-level verification. Experimental results demonstrate that StepProof significantly improves proof success rates and efficiency compared to traditional methods. Additionally, we found that minor manual adjustments to the natural language proofs, tailoring them for step-level verification, further enhanced StepProof's performance in autoformalization.
Authors: Yutong Zhou, Masahiro Ryo
Abstract: Explaining why the species lives at a particular location is important for understanding ecological systems and conserving biodiversity. However, existing ecological workflows are fragmented and often inaccessible to non-specialists. We propose an end-to-end visual-to-causal framework that transforms a species image into interpretable causal insights about its habitat preference. The system integrates species recognition, global occurrence retrieval, pseudo-absence sampling, and climate data extraction. We then discover causal structures among environmental features and estimate their influence on species occurrence using modern causal inference methods. Finally, we generate statistically grounded, human-readable causal explanations from structured templates and large language models. We demonstrate the framework on a bee and a flower species and report early results as part of an ongoing project, showing the potential of the multimodal AI assistant backed up by a recommended ecological modeling practice for describing species habitat in human-understandable language.
Authors: Imanol Solano, Julian Fierrez, Aythami Morales, Alejandro Pe\~na, Ruben Tolosana, Francisco Zamora-Martinez, Javier San Agustin
Abstract: Demographic bias in high-performance face recognition (FR) systems often eludes detection by existing metrics, especially with respect to subtle disparities in the tails of the score distribution. We introduce the Comprehensive Equity Index (CEI), a novel metric designed to address this limitation. CEI uniquely analyzes genuine and impostor score distributions separately, enabling a configurable focus on tail probabilities while also considering overall distribution shapes. Our extensive experiments (evaluating state-of-the-art FR systems, intentionally biased models, and diverse datasets) confirm CEI's superior ability to detect nuanced biases where previous methods fall short. Furthermore, we present CEI^A, an automated version of the metric that enhances objectivity and simplifies practical application. CEI provides a robust and sensitive tool for operational FR fairness assessment. The proposed methods have been developed particularly for bias evaluation in face biometrics but, in general, they are applicable for comparing statistical distributions in any problem where one is interested in analyzing the distribution tails.
Authors: Lizhen Wang, Zhurong Xia, Tianshu Hu, Pengrui Wang, Pengfei Wang, Zerong Zheng, Ming Zhou
Abstract: In e-commerce and digital marketing, generating high-fidelity human-product demonstration videos is important for effective product presentation. However, most existing frameworks either fail to preserve the identities of both humans and products or lack an understanding of human-product spatial relationships, leading to unrealistic representations and unnatural interactions. To address these challenges, we propose a Diffusion Transformer (DiT)-based framework. Our method simultaneously preserves human identities and product-specific details, such as logos and textures, by injecting paired human-product reference information and utilizing an additional masked cross-attention mechanism. We employ a 3D body mesh template and product bounding boxes to provide precise motion guidance, enabling intuitive alignment of hand gestures with product placements. Additionally, structured text encoding is used to incorporate category-level semantics, enhancing 3D consistency during small rotational changes across frames. Trained on a hybrid dataset with extensive data augmentation strategies, our approach outperforms state-of-the-art techniques in maintaining the identity integrity of both humans and products and generating realistic demonstration motions. Project page: https://submit2025-dream.github.io/DreamActor-H1/.
Authors: Antonio Ferrara, Francesco Cozzi, Alan Perotti, Andr\'e Panisson, Francesco Bonchi
Abstract: Determining whether an algorithmic decision-making system discriminates against a specific demographic typically involves comparing a single point estimate of a fairness metric against a predefined threshold. This practice is statistically brittle: it ignores sampling error and treats small demographic subgroups the same as large ones. The problem intensifies in intersectional analyses, where multiple sensitive attributes are considered jointly, giving rise to a larger number of smaller groups. As these groups become more granular, the data representing them becomes too sparse for reliable estimation, and fairness metrics yield excessively wide confidence intervals, precluding meaningful conclusions about potential unfair treatments. In this paper, we introduce a unified, size-adaptive, hypothesis-testing framework that turns fairness assessment into an evidence-based statistical decision. Our contribution is twofold. (i) For sufficiently large subgroups, we prove a Central-Limit result for the statistical parity difference, leading to analytic confidence intervals and a Wald test whose type-I (false positive) error is guaranteed at level $\alpha$. (ii) For the long tail of small intersectional groups, we derive a fully Bayesian Dirichlet-multinomial estimator; Monte-Carlo credible intervals are calibrated for any sample size and naturally converge to Wald intervals as more data becomes available. We validate our approach empirically on benchmark datasets, demonstrating how our tests provide interpretable, statistically rigorous decisions under varying degrees of data availability and intersectionality.
Authors: Xunguang Wang, Zhenlan Ji, Wenxuan Wang, Zongjie Li, Daoyuan Wu, Shuai Wang
Abstract: Large Language Models (LLMs) have achieved remarkable progress, but their deployment has exposed critical vulnerabilities, particularly to jailbreak attacks that circumvent safety mechanisms. Guardrails--external defense mechanisms that monitor and control LLM interaction--have emerged as a promising solution. However, the current landscape of LLM guardrails is fragmented, lacking a unified taxonomy and comprehensive evaluation framework. In this Systematization of Knowledge (SoK) paper, we present the first holistic analysis of jailbreak guardrails for LLMs. We propose a novel, multi-dimensional taxonomy that categorizes guardrails along six key dimensions, and introduce a Security-Efficiency-Utility evaluation framework to assess their practical effectiveness. Through extensive analysis and experiments, we identify the strengths and limitations of existing guardrail approaches, explore their universality across attack types, and provide insights into optimizing defense combinations. Our work offers a structured foundation for future research and development, aiming to guide the principled advancement and deployment of robust LLM guardrails. The code is available at https://github.com/xunguangwang/SoK4JailbreakGuardrails.
URLs: https://github.com/xunguangwang/SoK4JailbreakGuardrails.
Authors: Suin Lee, Dae-Shik Kim
Abstract: We present TexTailor, a novel method for generating consistent object textures from textual descriptions. Existing text-to-texture synthesis approaches utilize depth-aware diffusion models to progressively generate images and synthesize textures across predefined multiple viewpoints. However, these approaches lead to a gradual shift in texture properties across viewpoints due to (1) insufficient integration of previously synthesized textures at each viewpoint during the diffusion process and (2) the autoregressive nature of the texture synthesis process. Moreover, the predefined selection of camera positions, which does not account for the object's geometry, limits the effective use of texture information synthesized from different viewpoints, ultimately degrading overall texture consistency. In TexTailor, we address these issues by (1) applying a resampling scheme that repeatedly integrates information from previously synthesized textures within the diffusion process, and (2) fine-tuning a depth-aware diffusion model on these resampled textures. During this process, we observed that using only a few training images restricts the model's original ability to generate high-fidelity images aligned with the conditioning, and therefore propose an performance preservation loss to mitigate this issue. Additionally, we improve the synthesis of view-consistent textures by adaptively adjusting camera positions based on the object's geometry. Experiments on a subset of the Objaverse dataset and the ShapeNet car dataset demonstrate that TexTailor outperforms state-of-the-art methods in synthesizing view-consistent textures. The source code for TexTailor is available at https://github.com/Adios42/Textailor
Authors: Reza Karbasi, Masoud Rahimi, Abdol-Hossein Vahabie, Hadi Moradi
Abstract: This paper addresses the persistent challenge of accurately digitizing paper-based electrocardiogram (ECG) recordings, with a particular focus on robustly handling single leads compromised by signal overlaps-a common yet under-addressed issue in existing methodologies. We propose a two-stage pipeline designed to overcome this limitation. The first stage employs a U-Net based segmentation network, trained on a dataset enriched with overlapping signals and fortified with custom data augmentations, to accurately isolate the primary ECG trace. The subsequent stage converts this refined binary mask into a time-series signal using established digitization techniques, enhanced by an adaptive grid detection module for improved versatility across different ECG formats and scales. Our experimental results demonstrate the efficacy of our approach. The U-Net architecture achieves an IoU of 0.87 for the fine-grained segmentation task. Crucially, our proposed digitization method yields superior performance compared to a well-established baseline technique across both non-overlapping and challenging overlapping ECG samples. For non-overlapping signals, our method achieved a Mean Squared Error (MSE) of 0.0010 and a Pearson Correlation Coefficient (rho) of 0.9644, compared to 0.0015 and 0.9366, respectively, for the baseline. On samples with signal overlap, our method achieved an MSE of 0.0029 and a rho of 0.9641, significantly improving upon the baseline's 0.0178 and 0.8676. This work demonstrates an effective strategy to significantly enhance digitization accuracy, especially in the presence of signal overlaps, thereby laying a strong foundation for the reliable conversion of analog ECG records into analyzable digital data for contemporary research and clinical applications. The implementation is publicly available at this GitHub repository: https://github.com/masoudrahimi39/ECG-code.
Authors: Sergio Burdisso, Esa\'u Villatoro-Tello, Petr Motlicek
Abstract: The advancement of conversational AI systems relies on the availability of high-quality, flexible, and reproducible synthetic dialogues for training, evaluation, and benchmarking. SDialog is a modular, extensible Python toolkit designed to address the challenges of synthetic dialogue generation and analysis. By leveraging instruction-tuned Large Language Models (LLMs), SDialog provides abstractions for personas, orchestration, and scenario management, enabling the creation of realistic, diverse, and controllable conversational data for research and development. SDialog supports workflows such as multi-agent simulation and scenario-driven generation, and represents a step forward in the standardization of tools and frameworks for synthetic data generation, a crucial advancement for ensuring reproducibility in today's fast-evolving research landscape.
Authors: Numaan Naeem, Sarfraz Ahmad, Momina Ahsan, Hasan Iqbal
Abstract: This paper presents our system for Track 1: Mistake Identification in the BEA 2025 Shared Task on Pedagogical Ability Assessment of AI-powered Tutors. The task involves evaluating whether a tutor's response correctly identifies a mistake in a student's mathematical reasoning. We explore four approaches: (1) an ensemble of machine learning models over pooled token embeddings from multiple pretrained language models (LMs); (2) a frozen sentence-transformer using [CLS] embeddings with an MLP classifier; (3) a history-aware model with multi-head attention between token-level history and response embeddings; and (4) a retrieval-augmented few-shot prompting system with a large language model (LLM) i.e. GPT 4o. Our final system retrieves semantically similar examples, constructs structured prompts, and uses schema-guided output parsing to produce interpretable predictions. It outperforms all baselines, demonstrating the effectiveness of combining example-driven prompting with LLM reasoning for pedagogical feedback assessment. Our code is available at https://github.com/NaumanNaeem/BEA_2025.
Authors: Yucheng Yang, Tianyi Zhou, Qiang He, Lei Han, Mykola Pechenizkiy, Meng Fang
Abstract: Unsupervised reinforcement learning (URL) aims to learn general skills for unseen downstream tasks. Mutual Information Skill Learning (MISL) addresses URL by maximizing the mutual information between states and skills but lacks sufficient theoretical analysis, e.g., how well its learned skills can initialize a downstream task's policy. Our new theoretical analysis in this paper shows that the diversity and separability of learned skills are fundamentally critical to downstream task adaptation but MISL does not necessarily guarantee these properties. To complement MISL, we propose a novel disentanglement metric LSEPIN. Moreover, we build an information-geometric connection between LSEPIN and downstream task adaptation cost. For better geometric properties, we investigate a new strategy that replaces the KL divergence in information geometry with Wasserstein distance. We extend the geometric analysis to it, which leads to a novel skill-learning objective WSEP. It is theoretically justified to be helpful to downstream task adaptation and it is capable of discovering more initial policies for downstream tasks than MISL. We finally propose another Wasserstein distance-based algorithm PWSEP that can theoretically discover all optimal initial policies.
Authors: Yucong Luo, Yitong Zhou, Mingyue Cheng, Jiahao Wang, Daoyu Wang, Tingyue Pan, Jintao Zhang
Abstract: To advance time series forecasting (TSF), various methods have been proposed to improve prediction accuracy, evolving from statistical techniques to data-driven deep learning architectures. Despite their effectiveness, most existing methods still adhere to a fast thinking paradigm-relying on extracting historical patterns and mapping them to future values as their core modeling philosophy, lacking an explicit thinking process that incorporates intermediate time series reasoning. Meanwhile, emerging slow-thinking LLMs (e.g., OpenAI-o1) have shown remarkable multi-step reasoning capabilities, offering an alternative way to overcome these issues. However, prompt engineering alone presents several limitations - including high computational cost, privacy risks, and limited capacity for in-depth domain-specific time series reasoning. To address these limitations, a more promising approach is to train LLMs to develop slow thinking capabilities and acquire strong time series reasoning skills. For this purpose, we propose Time-R1, a two-stage reinforcement fine-tuning framework designed to enhance multi-step reasoning ability of LLMs for time series forecasting. Specifically, the first stage conducts supervised fine-tuning for warmup adaptation, while the second stage employs reinforcement learning to improve the model's generalization ability. Particularly, we design a fine-grained multi-objective reward specifically for time series forecasting, and then introduce GRIP (group-based relative importance for policy optimization), which leverages non-uniform sampling to further encourage and optimize the model's exploration of effective reasoning paths. Experiments demonstrate that Time-R1 significantly improves forecast performance across diverse datasets.
Authors: Francisco Caetano, Christiaan Viviers, Peter H. N. De With, Fons van der Sommen
Abstract: Flow Matching has emerged as a powerful framework for learning continuous transformations between distributions, enabling high-fidelity generative modeling. This work introduces Symmetrical Flow Matching (SymmFlow), a new formulation that unifies semantic segmentation, classification, and image generation within a single model. Using a symmetric learning objective, SymmFlow models forward and reverse transformations jointly, ensuring bi-directional consistency, while preserving sufficient entropy for generative diversity. A new training objective is introduced to explicitly retain semantic information across flows, featuring efficient sampling while preserving semantic structure, allowing for one-step segmentation and classification without iterative refinement. Unlike previous approaches that impose strict one-to-one mapping between masks and images, SymmFlow generalizes to flexible conditioning, supporting both pixel-level and image-level class labels. Experimental results on various benchmarks demonstrate that SymmFlow achieves state-of-the-art performance on semantic image synthesis, obtaining FID scores of 11.9 on CelebAMask-HQ and 7.0 on COCO-Stuff with only 25 inference steps. Additionally, it delivers competitive results on semantic segmentation and shows promising capabilities in classification tasks. The code will be publicly available.
Authors: Lang Yin, Debangshu Banerjee, Gagandeep Singh
Abstract: Chain of Thought (CoT) has been applied to various large language models (LLMs) and proven to be effective in improving the quality of outputs. In recent studies, transformers are proven to have absolute upper bounds in terms of expressive power, and consequently, they cannot solve many computationally difficult problems. However, empowered by CoT, transformers are proven to be able to solve some difficult problems effectively, such as the $k$-parity problem. Nevertheless, those works rely on two imperative assumptions: (1) identical training and testing distribution, and (2) corruption-free training data with correct reasoning steps. However, in the real world, these assumptions do not always hold. Although the risks of data shifts have caught attention, our work is the first to rigorously study the exact harm caused by such shifts to the best of our knowledge. Focusing on the $k$-parity problem, in this work we investigate the joint impact of two types of data shifts: the distribution shifts and data poisoning, on the quality of trained models obtained by a well-established CoT decomposition. In addition to revealing a surprising phenomenon that CoT leads to worse performance on learning parity than directly generating the prediction, our technical results also give a rigorous and comprehensive explanation of the mechanistic reasons of such impact.
Authors: Narges Nemati, Mostafa Haghir Chehreghani
Abstract: Matrix completion is a widely adopted framework in recommender systems, as predicting the missing entries in the user-item rating matrix enables a comprehensive understanding of user preferences. However, current graph neural network (GNN)-based approaches are highly sensitive to noisy or irrelevant edges--due to their inherent message-passing mechanisms--and are prone to overfitting, which limits their generalizability. To overcome these challenges, we propose a novel method called Matrix Completion using Contrastive Learning (MCCL). Our approach begins by extracting local neighborhood subgraphs for each interaction and subsequently generates two distinct graph representations. The first representation emphasizes denoising by integrating GNN layers with an attention mechanism, while the second is obtained via a graph variational autoencoder that aligns the feature distribution with a standard prior. A mutual learning loss function is employed during training to gradually harmonize these representations, enabling the model to capture common patterns and significantly enhance its generalizability. Extensive experiments on several real-world datasets demonstrate that our approach not only improves the numerical accuracy of the predicted scores--achieving up to a 0.8% improvement in RMSE--but also produces superior rankings with improvements of up to 36% in ranking metrics.
Authors: Marzieh Oghbaie, Teresa Ara\'ujoa, Hrvoje Bogunovi\'c
Abstract: Background and Objective: Prototype-based methods improve interpretability by learning fine-grained part-prototypes; however, their visualization in the input pixel space is not always consistent with human-understandable biomarkers. In addition, well-known prototype-based approaches typically learn extremely granular prototypes that are less interpretable in medical imaging, where both the presence and extent of biomarkers and lesions are critical. Methods: To address these challenges, we propose PiPViT (Patch-based Visual Interpretable Prototypes), an inherently interpretable prototypical model for image recognition. Leveraging a vision transformer (ViT), PiPViT captures long-range dependencies among patches to learn robust, human-interpretable prototypes that approximate lesion extent only using image-level labels. Additionally, PiPViT benefits from contrastive learning and multi-resolution input processing, which enables effective localization of biomarkers across scales. Results: We evaluated PiPViT on retinal OCT image classification across four datasets, where it achieved competitive quantitative performance compared to state-of-the-art methods while delivering more meaningful explanations. Moreover, quantitative evaluation on a hold-out test set confirms that the learned prototypes are semantically and clinically relevant. We believe PiPViT can transparently explain its decisions and assist clinicians in understanding diagnostic outcomes. Github page: https://github.com/marziehoghbaie/PiPViT
Authors: Igor Urbanik, Pawe{\l} Gajewski
Abstract: Continual learning poses a fundamental challenge for neural systems, which often suffer from catastrophic forgetting when exposed to sequential tasks. Self-Organizing Maps (SOMs), despite their interpretability and efficiency, are not immune to this issue. In this paper, we introduce Saturation Self-Organizing Maps (SatSOM)-an extension of SOMs designed to improve knowledge retention in continual learning scenarios. SatSOM incorporates a novel saturation mechanism that gradually reduces the learning rate and neighborhood radius of neurons as they accumulate information. This effectively freezes well-trained neurons and redirects learning to underutilized areas of the map.
Authors: Thanh Thi Nguyen, Campbell Wilson, Janis Dalins
Abstract: Detecting life-threatening language is essential for safeguarding individuals in distress, promoting mental health and well-being, and preventing potential harm and loss of life. This paper presents an effective approach to identifying life-threatening texts using large language models (LLMs) and compares them with traditional methods such as bag of words, word embedding, topic modeling, and Bidirectional Encoder Representations from Transformers. We fine-tune three open-source LLMs including Gemma, Mistral, and Llama-2 using their 7B parameter variants on different datasets, which are constructed with class balance, imbalance, and extreme imbalance scenarios. Experimental results demonstrate a strong performance of LLMs against traditional methods. More specifically, Mistral and Llama-2 models are top performers in both balanced and imbalanced data scenarios while Gemma is slightly behind. We employ the upsampling technique to deal with the imbalanced data scenarios and demonstrate that while this method benefits traditional approaches, it does not have as much impact on LLMs. This study demonstrates a great potential of LLMs for real-world life-threatening language detection problems.
Authors: Arshad Beg, Diarmuid O'Donoghue, Rosemary Monahan
Abstract: This paper is a brief introduction to our recently initiated project named VERIFAI: Traceability and verification of natural language requirements. The project addresses the challenges in the traceability and verification of formal specifications through providing support for the automatic generation of the formal specifications and the traceability of the requirements from the initial software design stage through the systems implementation and verification. Approaches explored in this project include Natural Language Processing, use of ontologies to describe the software system domain, reuse of existing software artefacts from similar systems (i.e. through similarity based reuse) and large language models to identify and declare the specifications as well as use of artificial intelligence to guide the process.
Authors: Marco Spinaci, Marek Polewczyk, Maximilian Schambach, Sam Thelin
Abstract: Tabular in-context learning (ICL) has recently achieved state-of-the-art (SOTA) performance on several tabular prediction tasks. Previously restricted to classification problems on small tables, recent advances such as TabPFN and TabICL have extended its use to larger datasets. While being architecturally efficient and well-adapted to tabular data structures, current table-native ICL architectures, being trained exclusively on synthetic data, do not fully leverage the rich semantics and world knowledge contained in real-world tabular data. On another end of this spectrum, tabular ICL models based on pretrained large language models such as TabuLa-8B integrate deep semantic understanding and world knowledge but are only able to make use of a small amount of context due to inherent architectural limitations. With the aim to combine the best of both these worlds, we introduce ConTextTab, integrating semantic understanding and alignment into a table-native ICL framework. By employing specialized embeddings for different data modalities and by training on large-scale real-world tabular data, our model is competitive with SOTA across a broad set of benchmarks while setting a new standard on the semantically rich CARTE benchmark.
Authors: Melika Ayoughi, Mina Ghadimi Atigh, Mohammad Mahdi Derakhshani, Cees G. M. Snoek, Pascal Mettes, Paul Groth
Abstract: Continual learning has traditionally focused on classifying either instances or classes, but real-world applications, such as robotics and self-driving cars, require models to handle both simultaneously. To mirror real-life scenarios, we introduce the task of continual learning of instances and classes, at the same time. This task challenges models to adapt to multiple levels of granularity over time, which requires balancing fine-grained instance recognition with coarse-grained class generalization. In this paper, we identify that classes and instances naturally form a hierarchical structure. To model these hierarchical relationships, we propose HyperCLIC, a continual learning algorithm that leverages hyperbolic space, which is uniquely suited for hierarchical data due to its ability to represent tree-like structures with low distortion and compact embeddings. Our framework incorporates hyperbolic classification and distillation objectives, enabling the continual embedding of hierarchical relations. To evaluate performance across multiple granularities, we introduce continual hierarchical metrics. We validate our approach on EgoObjects, the only dataset that captures the complexity of hierarchical object recognition in dynamic real-world environments. Empirical results show that HyperCLIC operates effectively at multiple granularities with improved hierarchical generalization.
Authors: Em\'ilio Dolgener Cant\'u, Rolf Klemens Wittmann, Oliver Abdeen, Patrick Wagner, Wojciech Samek, Moritz Baier, Sebastian Lapuschkin
Abstract: Quality management in semiconductor manufacturing often relies on template matching with known golden standards. For Indium-Phosphide (InP) multi-project wafer manufacturing, low production scale and high design variability lead to such golden standards being typically unavailable. Defect detection, in turn, is manual and labor-intensive. This work addresses this challenge by proposing a methodology to generate a synthetic golden standard using Deep Neural Networks, trained to simulate photo-realistic InP wafer images from CAD data. We evaluate various training objectives and assess the quality of the simulated images on both synthetic data and InP wafer photographs. Our deep-learning-based method outperforms a baseline decision-tree-based approach, enabling the use of a 'simulated golden die' from CAD plans in any user-defined region of a wafer for more efficient defect detection. We apply our method to a template matching procedure, to demonstrate its practical utility in surface defect detection.
Authors: Ye Yu, Yaoning Yu, Haohan Wang
Abstract: Large reasoning models (LRMs) such as Claude 3.7 Sonnet and OpenAI o1 achieve strong performance on mathematical benchmarks using lengthy chain-of-thought (CoT) reasoning, but the resulting traces are often unnecessarily verbose. This inflates token usage and cost, limiting deployment in latency-sensitive or API-constrained settings. We introduce PREMISE (PRompt-based Efficient Mathematical Inference with Strategic Evaluation), a prompt-only framework that reduces reasoning overhead without modifying model weights. PREMISE combines trace-level diagnostics with gradient-inspired prompt optimization to minimize redundant computation while preserving answer accuracy. The approach jointly optimizes brevity and correctness through a multi-objective textual search that balances token length and answer validity. Unlike prior work, PREMISE runs in a single-pass black-box interface, so it can be applied directly to commercial LLMs. On GSM8K, SVAMP, and Math500 we match or exceed baseline accuracy ($96\%\rightarrow96\%$ with Claude, $91\%\rightarrow92\%$ with Gemini) while reducing reasoning tokens by up to $87.5\%$ and cutting dollar cost by $69$--$82\%$. These results show that prompt-level optimization is a practical and scalable path to efficient LRM inference without compromising reasoning quality.
Authors: Xiaoxing Mo, Yuxuan Cheng, Nan Sun, Leo Yu Zhang, Wei Luo, Shang Gao
Abstract: Deep Neural Networks (DNNs) are vulnerable to backdoor attacks, where attackers implant hidden triggers during training to maliciously control model behavior. Topological Evolution Dynamics (TED) has recently emerged as a powerful tool for detecting backdoor attacks in DNNs. However, TED can be vulnerable to backdoor attacks that adaptively distort topological representation distributions across network layers. To address this limitation, we propose TED-LaST (Topological Evolution Dynamics against Laundry, Slow release, and Target mapping attack strategies), a novel defense strategy that enhances TED's robustness against adaptive attacks. TED-LaST introduces two key innovations: label-supervised dynamics tracking and adaptive layer emphasis. These enhancements enable the identification of stealthy threats that evade traditional TED-based defenses, even in cases of inseparability in topological space and subtle topological perturbations. We review and classify data poisoning tricks in state-of-the-art adaptive attacks and propose enhanced adaptive attack with target mapping, which can dynamically shift malicious tasks and fully leverage the stealthiness that adaptive attacks possess. Our comprehensive experiments on multiple datasets (CIFAR-10, GTSRB, and ImageNet100) and model architectures (ResNet20, ResNet101) show that TED-LaST effectively counteracts sophisticated backdoors like Adap-Blend, Adapt-Patch, and the proposed enhanced adaptive attack. TED-LaST sets a new benchmark for robust backdoor detection, substantially enhancing DNN security against evolving threats.
Authors: Chi Zuo, Martin B. M{\o}ller, Pablo Mart\'inez-Nuevo, Huayang Huang, Yu Wu, Ye Zhu
Abstract: While being disturbed by environmental noises, the acoustic masking technique is a conventional way to reduce the annoyance in audio engineering that seeks to cover up the noises with other dominant yet less intrusive sounds. However, misalignment between the dominant sound and the noise-such as mismatched downbeats-often requires an excessive volume increase to achieve effective masking. Motivated by recent advances in cross-modal generation, in this work, we introduce an alternative method to acoustic masking, aiming to reduce the noticeability of environmental noises by blending them into personalized music generated based on user-provided text prompts. Following the paradigm of music generation using mel-spectrogram representations, we propose a Blending Noises into Personalized Music (BNMusic) framework with two key stages. The first stage synthesizes a complete piece of music in a mel-spectrogram representation that encapsulates the musical essence of the noise. In the second stage, we adaptively amplify the generated music segment to further reduce noise perception and enhance the blending effectiveness, while preserving auditory quality. Our experiments with comprehensive evaluations on MusicBench, EPIC-SOUNDS, and ESC-50 demonstrate the effectiveness of our framework, highlighting the ability to blend environmental noise with rhythmically aligned, adaptively amplified, and enjoyable music segments, minimizing the noticeability of the noise, thereby improving overall acoustic experiences.
Authors: Yuhang Zhang, Haosheng Yu, Jiaping Xiao, Mir Feroskhan
Abstract: Vision-and-language navigation (VLN) is a long-standing challenge in autonomous robotics, aiming to empower agents with the ability to follow human instructions while navigating complex environments. Two key bottlenecks remain in this field: generalization to out-of-distribution environments and reliance on fixed discrete action spaces. To address these challenges, we propose Vision-Language Fly (VLFly), a framework tailored for Unmanned Aerial Vehicles (UAVs) to execute language-guided flight. Without the requirement for localization or active ranging sensors, VLFly outputs continuous velocity commands purely from egocentric observations captured by an onboard monocular camera. The VLFly integrates three modules: an instruction encoder based on a large language model (LLM) that reformulates high-level language into structured prompts, a goal retriever powered by a vision-language model (VLM) that matches these prompts to goal images via vision-language similarity, and a waypoint planner that generates executable trajectories for real-time UAV control. VLFly is evaluated across diverse simulation environments without additional fine-tuning and consistently outperforms all baselines. Moreover, real-world VLN tasks in indoor and outdoor environments under direct and indirect instructions demonstrate that VLFly achieves robust open-vocabulary goal understanding and generalized navigation capabilities, even in the presence of abstract language input.
Authors: Yinhao Xu, Georg A. Gottwald, Zdenka Kuncic
Abstract: This study investigates how dynamical systems may be learned and modelled with a neuromorphic network which is itself a dynamical system. The neuromorphic network used in this study is based on a complex electrical circuit comprised of memristive elements that produce neuro-synaptic nonlinear responses to input electrical signals. To determine how computation may be performed using the physics of the underlying system, the neuromorphic network was simulated and evaluated on autonomous prediction of a multivariate chaotic time series, implemented with a reservoir computing framework. Through manipulating only input electrodes and voltages, optimal nonlinear dynamical responses were found when input voltages maximise the number of memristive components whose internal dynamics explore the entire dynamical range of the memristor model. Increasing the network coverage with the input electrodes was found to suppress other nonlinear responses that are less conducive to learning. These results provide valuable insights into how a practical neuromorphic network device can be optimised for learning complex dynamical systems using only external control parameters.
Authors: Wenhao Guo, Peng Lu, Xujun Peng, Zhaoran Zhao, Sheng Li
Abstract: Prior Arbitrary-Scale Image Super-Resolution (ASISR) methods often experience a significant performance decline when the upsampling factor exceeds the range covered by the training data, introducing substantial blurring. To address this issue, we propose a unified model, Stroke-based Cyclic Amplifier (SbCA), for ultra-large upsampling tasks. The key of SbCA is the stroke vector amplifier, which decomposes the image into a series of strokes represented as vector graphics for magnification. Then, the detail completion module also restores missing details, ensuring high-fidelity image reconstruction. Our cyclic strategy achieves ultra-large upsampling by iteratively refining details with this unified SbCA model, trained only once for all, while keeping sub-scales within the training range. Our approach effectively addresses the distribution drift issue and eliminates artifacts, noise and blurring, producing high-quality, high-resolution super-resolved images. Experimental validations on both synthetic and real-world datasets demonstrate that our approach significantly outperforms existing methods in ultra-large upsampling tasks (e.g. $\times100$), delivering visual quality far superior to state-of-the-art techniques.
Authors: Feiyu Yang, Siyuan Liang, Aishan Liu, Dacheng Tao
Abstract: The capability of generative diffusion models (DMs) like Stable Diffusion (SD) in replicating training data could be taken advantage of by attackers to launch the Copyright Infringement Attack, with duplicated poisoned image-text pairs. SilentBadDiffusion (SBD) is a method proposed recently, which shew outstanding performance in attacking SD in text-to-image tasks. However, the feasible data resources in this area are still limited, some of them are even constrained or prohibited due to the issues like copyright ownership or inappropriate contents; And not all of the images in current datasets are suitable for the proposed attacking methods; Besides, the state-of-the-art (SoTA) performance of SBD is far from ideal when few generated poisoning samples could be adopted for attacks. In this paper, we raised new datasets accessible for researching in attacks like SBD, and proposed Multi-Element (ME) attack method based on SBD by increasing the number of poisonous visual-text elements per poisoned sample to enhance the ability of attacking, while importing Discrete Cosine Transform (DCT) for the poisoned samples to maintain the stealthiness. The Copyright Infringement Rate (CIR) / First Attack Epoch (FAE) we got on the two new datasets were 16.78% / 39.50 and 51.20% / 23.60, respectively close to or even outperformed benchmark Pokemon and Mijourney datasets. In condition of low subsampling ratio (5%, 6 poisoned samples), MESI and DCT earned CIR / FAE of 0.23% / 84.00 and 12.73% / 65.50, both better than original SBD, which failed to attack at all.
Authors: Jian Li, Wan Han, Ning Lin, Yu-Liang Zhan, Ruizhi Chengze, Haining Wang, Yi Zhang, Hongsheng Liu, Zidong Wang, Fan Yu, Hao Sun
Abstract: Understanding and reasoning about dynamics governed by physical laws through visual observation, akin to human capabilities in the real world, poses significant challenges. Currently, object-centric dynamic simulation methods, which emulate human behavior, have achieved notable progress but overlook two critical aspects: 1) the integration of physical knowledge into models. Humans gain physical insights by observing the world and apply this knowledge to accurately reason about various dynamic scenarios; 2) the validation of model adaptability across diverse scenarios. Real-world dynamics, especially those involving fluids and objects, demand models that not only capture object interactions but also simulate fluid flow characteristics. To address these gaps, we introduce SlotPi, a slot-based physics-informed object-centric reasoning model. SlotPi integrates a physical module based on Hamiltonian principles with a spatio-temporal prediction module for dynamic forecasting. Our experiments highlight the model's strengths in tasks such as prediction and Visual Question Answering (VQA) on benchmark and fluid datasets. Furthermore, we have created a real-world dataset encompassing object interactions, fluid dynamics, and fluid-object interactions, on which we validated our model's capabilities. The model's robust performance across all datasets underscores its strong adaptability, laying a foundation for developing more advanced world models.
Authors: Viet Anh Trinh, Xinlu He, Jacob Whitehill
Abstract: With recent advances in modeling and the increasing amount of supervised training data, automatic speech recognition (ASR) systems have achieved remarkable performance on general speech. However, the word error rate (WER) of state-of-the-art ASR remains high for named entities. Since named entities are often the most critical keywords, misrecognizing them can affect all downstream applications, especially when the ASR system functions as the front end of a complex system. In this paper, we introduce a large language model (LLM) revision mechanism to revise incorrect named entities in ASR predictions by leveraging the LLM's reasoning ability as well as local context (e.g., lecture notes) containing a set of correct named entities. Finally, we introduce the NER-MIT-OpenCourseWare dataset, containing 45 hours of data from MIT courses for development and testing. On this dataset, our proposed technique achieves up to 30\% relative WER reduction for named entities.
Authors: Vinaik Chhetri, Krishna Upadhyay, A. B. Siddique, Umar Farooq
Abstract: Artificial Intelligence (AI)-powered features have rapidly proliferated across mobile apps in various domains, including productivity, education, entertainment, and creativity. However, how users perceive, evaluate, and critique these AI features remains largely unexplored, primarily due to the overwhelming volume of user feedback. In this work, we present the first comprehensive, large-scale study of user feedback on AI-powered mobile apps, leveraging a curated dataset of 292 AI-driven apps across 14 categories with 894K AI-specific reviews from Google Play. We develop and validate a multi-stage analysis pipeline that begins with a human-labeled benchmark and systematically evaluates large language models (LLMs) and prompting strategies. Each stage, including review classification, aspect-sentiment extraction, and clustering, is validated for accuracy and consistency. Our pipeline enables scalable, high-precision analysis of user feedback, extracting over one million aspect-sentiment pairs clustered into 18 positive and 15 negative user topics. Our analysis reveals that users consistently focus on a narrow set of themes: positive comments emphasize productivity, reliability, and personalized assistance, while negative feedback highlights technical failures (e.g., scanning and recognition), pricing concerns, and limitations in language support. Our pipeline surfaces both satisfaction with one feature and frustration with another within the same review. These fine-grained, co-occurring sentiments are often missed by traditional approaches that treat positive and negative feedback in isolation or rely on coarse-grained analysis. To this end, our approach provides a more faithful reflection of the real-world user experiences with AI-powered apps. Category-aware analysis further uncovers both universal drivers of satisfaction and domain-specific frustrations.
Authors: Huaying Yuan, Zheng Liu, Junjie Zhou, Ji-Rong Wen, Zhicheng Dou
Abstract: Long video understanding (LVU) presents a significant challenge for current multi-modal large language models (MLLMs) due to the task's inherent complexity and context window constraint. It is widely assumed that addressing LVU tasks requires foundation MLLMs with extended context windows, strong visual perception capabilities, and proficient domain expertise. In this work, we challenge this common belief by introducing VideoDeepResearch, a novel agentic framework for long video understanding. Our approach relies solely on a text-only large reasoning model (LRM) combined with a modular multi-modal toolkit, including multimodal retrievers and visual perceivers, all of which are readily available in practice. For each LVU task, the system formulates a problem-solving strategy through reasoning, while selectively accessing and utilizing essential video content via tool using. We conduct extensive experiments on popular LVU benchmarks, including MLVU, Video-MME, and LVBench. Our results demonstrate that VideoDeepResearch achieves substantial improvements over existing MLLM baselines, surpassing the previous state-of-the-art by 9.6%, 6.6%, and 3.9% on MLVU (test), LVBench, and LongVideoBench, respectively. These findings highlight the promise of agentic systems in overcoming key challenges in LVU problems.
Authors: Andrea Moglia (Politecnico di Milano), Matteo Leccardi (Politecnico di Milano), Matteo Cavicchioli (Politecnico di Milano), Alice Maccarini (Universit\`a di Pavia), Marco Marcon (Politecnico di Milano), Luca Mainardi (Politecnico di Milano), Pietro Cerveri (Politecnico di Milano, Universit\`a di Pavia)
Abstract: Following the successful paradigm shift of large language models, leveraging pre-training on a massive corpus of data and fine-tuning on different downstream tasks, generalist models have made their foray into computer vision. The introduction of Segment Anything Model (SAM) set a milestone on segmentation of natural images, inspiring the design of a multitude of architectures for medical image segmentation. In this survey we offer a comprehensive and in-depth investigation on generalist models for medical image segmentation. We start with an introduction on the fundamentals concepts underpinning their development. Then, we provide a taxonomy on the different declinations of SAM in terms of zero-shot, few-shot, fine-tuning, adapters, on the recent SAM 2, on other innovative models trained on images alone, and others trained on both text and images. We thoroughly analyze their performances at the level of both primary research and best-in-literature, followed by a rigorous comparison with the state-of-the-art task-specific models. We emphasize the need to address challenges in terms of compliance with regulatory frameworks, privacy and security laws, budget, and trustworthy artificial intelligence (AI). Finally, we share our perspective on future directions concerning synthetic data, early fusion, lessons learnt from generalist models in natural language processing, agentic AI and physical AI, and clinical translation.
Authors: Mohammadreza Molavi, Mohammadreza Tavakoli, Mohammad Moein, Abdolali Faraji, G\'abor Kismih\'ok
Abstract: Online learning has experienced rapid growth due to its flexibility and accessibility. Personalization, adapted to the needs of individual learners, is crucial for enhancing the learning experience, particularly in online settings. A key aspect of personalization is providing learners with answers customized to their specific questions. This paper therefore explores the potential of Large Language Models (LLMs) to generate personalized answers to learners' questions, thereby enhancing engagement and reducing the workload on educators. To evaluate the effectiveness of LLMs in this context, we conducted a comprehensive study using the StackExchange platform in two distinct areas: language learning and programming. We developed a framework and a dataset for validating automatically generated personalized answers. Subsequently, we generated personalized answers using different strategies, including 0-shot, 1-shot, and few-shot scenarios. The generated answers were evaluated using three methods: 1. BERTScore, 2. LLM evaluation, and 3. human evaluation. Our findings indicated that providing LLMs with examples of desired answers (from the learner or similar learners) can significantly enhance the LLMs' ability to tailor responses to individual learners' needs.
Authors: Ravishka Rathnasuriya, Tingxi Li, Zexin Xu, Zihe Song, Mirazul Haque, Simin Chen, Wei Yang
Abstract: Deep Learning Systems (DLSs) are increasingly deployed in real-time applications, including those in resourceconstrained environments such as mobile and IoT devices. To address efficiency challenges, Dynamic Deep Learning Systems (DDLSs) adapt inference computation based on input complexity, reducing overhead. While this dynamic behavior improves efficiency, such behavior introduces new attack surfaces. In particular, efficiency adversarial attacks exploit these dynamic mechanisms to degrade system performance. This paper systematically explores efficiency robustness of DDLSs, presenting the first comprehensive taxonomy of efficiency attacks. We categorize these attacks based on three dynamic behaviors: (i) attacks on dynamic computations per inference, (ii) attacks on dynamic inference iterations, and (iii) attacks on dynamic output production for downstream tasks. Through an in-depth evaluation, we analyze adversarial strategies that target DDLSs efficiency and identify key challenges in securing these systems. In addition, we investigate existing defense mechanisms, demonstrating their limitations against increasingly popular efficiency attacks and the necessity for novel mitigation strategies to secure future adaptive DDLSs.
Authors: Tianrui Zhu, Houyuan Chen, Ruihao Gong, Michele Magno, Haotong Qin, Kai Zhang
Abstract: Video matting is crucial for applications such as film production and virtual reality, yet deploying its computationally intensive models on resource-constrained devices presents challenges. Quantization is a key technique for model compression and acceleration. As an efficient approach, Post-Training Quantization (PTQ) is still in its nascent stages for video matting, facing significant hurdles in maintaining accuracy and temporal coherence. To address these challenges, this paper proposes a novel and general PTQ framework specifically designed for video matting models, marking, to the best of our knowledge, the first systematic attempt in this domain. Our contributions include: (1) A two-stage PTQ strategy that combines block-reconstruction-based optimization for fast, stable initial quantization and local dependency capture, followed by a global calibration of quantization parameters to minimize accuracy loss. (2) A Statistically-Driven Global Affine Calibration (GAC) method that enables the network to compensate for cumulative statistical distortions arising from factors such as neglected BN layer effects, even reducing the error of existing PTQ methods on video matting tasks up to 20%. (3) An Optical Flow Assistance (OFA) component that leverages temporal and semantic priors from frames to guide the PTQ process, enhancing the model's ability to distinguish moving foregrounds in complex scenes and ultimately achieving near full-precision performance even under ultra-low-bit quantization. Comprehensive quantitative and visual results show that our PTQ4VM achieves the state-of-the-art accuracy performance across different bit-widths compared to the existing quantization methods. We highlight that the 4-bit PTQ4VM even achieves performance close to the full-precision counterpart while enjoying 8x FLOP savings.
Authors: Qingyan Wei, Yaojie Zhang, Zhiyuan Liu, Dongrui Liu, Linfeng Zhang
Abstract: Diffusion-based language models (dLLMs) have emerged as a promising alternative to traditional autoregressive LLMs by enabling parallel token generation and significantly reducing inference latency. However, existing sampling strategies for dLLMs, such as confidence-based or semi-autoregressive decoding, often suffer from static behavior, leading to suboptimal efficiency and limited flexibility. In this paper, we propose SlowFast Sampling, a novel dynamic sampling strategy that adaptively alternates between exploratory and accelerated decoding stages. Our method is guided by three golden principles: certainty principle, convergence principle, and positional principle, which govern when and where tokens can be confidently and efficiently decoded. We further integrate our strategy with dLLM-Cache to reduce redundant computation. Extensive experiments across benchmarks and models show that SlowFast Sampling achieves up to 15.63$\times$ speedup on LLaDA with minimal accuracy drop, and up to 34.22$\times$ when combined with caching. Notably, our approach outperforms strong autoregressive baselines like LLaMA3 8B in throughput, demonstrating that well-designed sampling can unlock the full potential of dLLMs for fast and high-quality generation.
Authors: Jiashuo Yu, Yue Wu, Meng Chu, Zhifei Ren, Zizheng Huang, Pei Chu, Ruijie Zhang, Yinan He, Qirui Li, Songze Li, Zhenxiang Li, Zhongying Tu, Conghui He, Yu Qiao, Yali Wang, Yi Wang, Limin Wang
Abstract: We present VRBench, the first long narrative video benchmark crafted for evaluating large models' multi-step reasoning capabilities, addressing limitations in existing evaluations that overlook temporal reasoning and procedural validity. It comprises 1,010 long videos (with an average duration of 1.6 hours), along with 9,468 human-labeled multi-step question-answering pairs and 30,292 reasoning steps with timestamps. These videos are curated via a multi-stage filtering process including expert inter-rater reviewing to prioritize plot coherence. We develop a human-AI collaborative framework that generates coherent reasoning chains, each requiring multiple temporally grounded steps, spanning seven types (e.g., event attribution, implicit inference). VRBench designs a multi-phase evaluation pipeline that assesses models at both the outcome and process levels. Apart from the MCQs for the final results, we propose a progress-level LLM-guided scoring metric to evaluate the quality of the reasoning chain from multiple dimensions comprehensively. Through extensive evaluations of 12 LLMs and 16 VLMs on VRBench, we undertake a thorough analysis and provide valuable insights that advance the field of multi-step reasoning.
Authors: Kehan Long, Shasha Li, Chen Xu, Jintao Tang, Ting Wang
Abstract: Recent advancements have successfully harnessed the power of Large Language Models (LLMs) for zero-shot document ranking, exploring a variety of prompting strategies. Comparative approaches like pairwise and listwise achieve high effectiveness but are computationally intensive and thus less practical for larger-scale applications. Scoring-based pointwise approaches exhibit superior efficiency by independently and simultaneously generating the relevance scores for each candidate document. However, this independence ignores critical comparative insights between documents, resulting in inconsistent scoring and suboptimal performance. In this paper, we aim to improve the effectiveness of pointwise methods while preserving their efficiency through two key innovations: (1) We propose a novel Global-Consistent Comparative Pointwise Ranking (GCCP) strategy that incorporates global reference comparisons between each candidate and an anchor document to generate contrastive relevance scores. We strategically design the anchor document as a query-focused summary of pseudo-relevant candidates, which serves as an effective reference point by capturing the global context for document comparison. (2) These contrastive relevance scores can be efficiently Post-Aggregated with existing pointwise methods, seamlessly integrating essential Global Context information in a training-free manner (PAGC). Extensive experiments on the TREC DL and BEIR benchmark demonstrate that our approach significantly outperforms previous pointwise methods while maintaining comparable efficiency. Our method also achieves competitive performance against comparative methods that require substantially more computational resources. More analyses further validate the efficacy of our anchor construction strategy.
Authors: Simon Lang, Martin Leutbecher, Pedro Maciel
Abstract: We assess the impact of a multi-scale loss formulation for training probabilistic machine-learned weather forecasting models. The multi-scale loss is tested in AIFS-CRPS, a machine-learned weather forecasting model developed at the European Centre for Medium-Range Weather Forecasts (ECMWF). AIFS-CRPS is trained by directly optimising the almost fair continuous ranked probability score (afCRPS). The multi-scale loss better constrains small scale variability without negatively impacting forecast skill. This opens up promising directions for future work in scale-aware model training.
Authors: Guanjin Wang, Xiangxue Zhao, Shapour Azarm, Balakumar Balachandran
Abstract: An alternative data-driven modeling approach has been proposed and employed to gain fundamental insights into robot motion interaction with granular terrain at certain length scales. The approach is based on an integration of dimension reduction (Sequentially Truncated Higher-Order Singular Value Decomposition), surrogate modeling (Gaussian Process), and data assimilation techniques (Reduced Order Particle Filter). This approach can be used online and is based on offline data, obtained from the offline collection of high-fidelity simulation data and a set of sparse experimental data. The results have shown that orders of magnitude reduction in computational time can be obtained from the proposed data-driven modeling approach compared with physics-based high-fidelity simulations. With only simulation data as input, the data-driven prediction technique can generate predictions that have comparable accuracy as simulations. With both simulation data and sparse physical experimental measurement as input, the data-driven approach with its embedded data assimilation techniques has the potential in outperforming only high-fidelity simulations for the long-horizon predictions. In addition, it is demonstrated that the data-driven modeling approach can also reproduce the scaling relationship recovered by physics-based simulations for maximum resistive forces, which may indicate its general predictability beyond a case-by-case basis. The results are expected to help robot navigation and exploration in unknown and complex terrains during both online and offline phases.
Authors: Qingda (Michael), Mai
Abstract: This paper investigates and validates the impact of fine-tuning on large language model performance, focusing on parameter-efficient methods (LoRA and QLoRA). We evaluate model capabilities across three key domains: (1) commonsense reasoning (HellaSwag), (2) mathematical reasoning (GSM8K), and (3) multi-domain knowledge (MMLU-CS). Our findings demonstrate that: (1) LoRA-based methods effectively improve task-specific performance while maintaining computational efficiency, and (2) performance strongly depends on alignment between fine-tuning dataset and benchmark tasks. The study provides both theoretical insights into parameter-efficient mechanisms and practical guidance for developers implementing efficient LLM adaptation with limited resources.
Authors: Subham Sekhar Sahoo, Justin Deschenaux, Aaron Gokaslan, Guanghan Wang, Justin Chiu, Volodymyr Kuleshov
Abstract: Uniform-state discrete diffusion models hold the promise of fast text generation due to their inherent ability to self-correct. However, they are typically outperformed by autoregressive models and masked diffusion models. In this work, we narrow this performance gap by leveraging a key insight: Uniform-state diffusion processes naturally emerge from an underlying Gaussian diffusion. Our method, Duo, transfers powerful techniques from Gaussian diffusion to improve both training and sampling. First, we introduce a curriculum learning strategy guided by the Gaussian process, doubling training speed by reducing variance. Models trained with curriculum learning surpass autoregressive models in zero-shot perplexity on 3 of 7 benchmarks. Second, we present Discrete Consistency Distillation, which adapts consistency distillation from the continuous to the discrete setting. This algorithm unlocks few-step generation in diffusion language models by accelerating sampling by two orders of magnitude. We provide the code and model checkpoints on the project page: http://s-sahoo.github.io/duo
Authors: Guimeng Liu, Milad Abdollahzadeh, Ngai-Man Cheung
Abstract: Zero-shot generative model adaptation (ZSGM) aims to adapt a pre-trained generator to a target domain using only text guidance and without any samples from the target domain. Central to recent ZSGM approaches are directional loss which use the text guidance in the form of aligning the image offset with text offset in the embedding space of a vision-language model like CLIP. This is similar to the analogical reasoning in NLP where the offset between one pair of words is used to identify a missing element in another pair by aligning the offset between these two pairs. However, a major limitation of existing ZSGM methods is that the learning objective assumes the complete alignment between image offset and text offset in the CLIP embedding space, resulting in quality degrade in generated images. Our work makes two main contributions. Inspired by the offset misalignment studies in NLP, as our first contribution, we perform an empirical study to analyze the misalignment between text offset and image offset in CLIP embedding space for various large publicly available datasets. Our important finding is that offset misalignment in CLIP embedding space is correlated with concept distance, i.e., close concepts have a less offset misalignment. To address the limitations of the current approaches, as our second contribution, we propose Adaptation with Iterative Refinement (AIR) which is the first ZSGM approach to focus on improving target domain image quality based on our new insight on offset misalignment.Qualitative, quantitative, and user study in 26 experiment setups consistently demonstrate the proposed AIR approach achieves SOTA performance. Additional experiments are in Supp.
Authors: Thomas Sounack, Joshua Davis, Brigitte Durieux, Antoine Chaffin, Tom J. Pollard, Eric Lehman, Alistair E. W. Johnson, Matthew McDermott, Tristan Naumann, Charlotta Lindvall
Abstract: Encoder-based transformer models are central to biomedical and clinical Natural Language Processing (NLP), as their bidirectional self-attention makes them well-suited for efficiently extracting structured information from unstructured text through discriminative tasks. However, encoders have seen slower development compared to decoder models, leading to limited domain adaptation in biomedical and clinical settings. We introduce BioClinical ModernBERT, a domain-adapted encoder that builds on the recent ModernBERT release, incorporating long-context processing and substantial improvements in speed and performance for biomedical and clinical NLP. BioClinical ModernBERT is developed through continued pretraining on the largest biomedical and clinical corpus to date, with over 53.5 billion tokens, and addresses a key limitation of prior clinical encoders by leveraging 20 datasets from diverse institutions, domains, and geographic regions, rather than relying on data from a single source. It outperforms existing biomedical and clinical encoders on four downstream tasks spanning a broad range of use cases. We release both base (150M parameters) and large (396M parameters) versions of BioClinical ModernBERT, along with training checkpoints to support further research.
Authors: Jiancheng Huang, Gengwei Zhang, Zequn Jie, Siyu Jiao, Yinlong Qian, Ling Chen, Yunchao Wei, Lin Ma
Abstract: Text-to-video generation has significantly enriched content creation and holds the potential to evolve into powerful world simulators. However, modeling the vast spatiotemporal space remains computationally demanding, particularly when employing Transformers, which incur quadratic complexity in sequence processing and thus limit practical applications. Recent advancements in linear-time sequence modeling, particularly the Mamba architecture, offer a more efficient alternative. Nevertheless, its plain design limits its direct applicability to multi-modal and spatiotemporal video generation tasks. To address these challenges, we introduce M4V, a Multi-Modal Mamba framework for text-to-video generation. Specifically, we propose a multi-modal diffusion Mamba (MM-DiM) block that enables seamless integration of multi-modal information and spatiotemporal modeling through a multi-modal token re-composition design. As a result, the Mamba blocks in M4V reduce FLOPs by 45% compared to the attention-based alternative when generating videos at 768$\times$1280 resolution. Additionally, to mitigate the visual quality degradation in long-context autoregressive generation processes, we introduce a reward learning strategy that further enhances per-frame visual realism. Extensive experiments on text-to-video benchmarks demonstrate M4V's ability to produce high-quality videos while significantly lowering computational costs. Code and models will be publicly available at https://huangjch526.github.io/M4V_project.
Authors: Adam Karvonen, Samuel Marks
Abstract: Large language models (LLMs) are increasingly deployed in high-stakes hiring applications, making decisions that directly impact people's careers and livelihoods. While prior studies suggest simple anti-bias prompts can eliminate demographic biases in controlled evaluations, we find these mitigations fail when realistic contextual details are introduced. We address these failures through internal bias mitigation: by identifying and neutralizing sensitive attribute directions within model activations, we achieve robust bias reduction across all tested scenarios. Across leading commercial (GPT-4o, Claude 4 Sonnet, Gemini 2.5 Flash) and open-source models (Gemma-2 27B, Gemma-3, Mistral-24B), we find that adding realistic context such as company names, culture descriptions from public careers pages, and selective hiring constraints (e.g.,``only accept candidates in the top 10\%") induces significant racial and gender biases (up to 12\% differences in interview rates). When these biases emerge, they consistently favor Black over White candidates and female over male candidates across all tested models and scenarios. Moreover, models can infer demographics and become biased from subtle cues like college affiliations, with these biases remaining invisible even when inspecting the model's chain-of-thought reasoning. To address these limitations, our internal bias mitigation identifies race and gender-correlated directions and applies affine concept editing at inference time. Despite using directions from a simple synthetic dataset, the intervention generalizes robustly, consistently reducing bias to very low levels (typically under 1\%, always below 2.5\%) while largely maintaining model performance. Our findings suggest that practitioners deploying LLMs for hiring should adopt more realistic evaluation methodologies and consider internal mitigation strategies for equitable outcomes.
Authors: Eduardo Baena, Paolo Testolina, Michele Polese, Sergi Aliaga, Andrew Benincasa, Dimitrios Koutsonikolas, Josep Jornet, Tommaso Melodia
Abstract: Lunar surface operations impose stringent requirements on wireless communication systems, including autonomy, robustness to disruption, and the ability to adapt to environmental and mission-driven context. While Space-O-RAN provides a distributed orchestration model aligned with 3GPP standards, its decision logic is limited to static policies and lacks semantic integration. We propose a novel extension incorporating a semantic agentic layer enabled by the Model Context Protocol (MCP) and Agent-to-Agent (A2A) communication protocols, allowing context-aware decision making across real-time, near-real-time, and non-real-time control layers. Distributed cognitive agents deployed in rovers, landers, and lunar base stations implement wireless-aware coordination strategies, including delay-adaptive reasoning and bandwidth-aware semantic compression, while interacting with multiple MCP servers to reason over telemetry, locomotion planning, and mission constraints.
Authors: T. T. J. E. Arets, G. Perugia, M. Houben, W. A. IJsselsteijn
Abstract: Reduced social connectedness increasingly poses a threat to mental health, life expectancy, and general well-being. Generative AI (GAI) technologies, such as large language models (LLMs) and image generation tools, are increasingly integrated into applications aimed at enhancing human social experiences. Despite their growing presence, little is known about how these technologies influence social interactions. This scoping review investigates how GAI-based applications are currently designed to facilitate social interaction, what forms of social engagement they target, and which design and evaluation methodologies designers use to create and evaluate them. Through an analysis of 30 studies published since 2020, we identify key trends in application domains including storytelling, socio-emotional skills training, reminiscence, collaborative learning, music making, and general conversation. We highlight the role of participatory and co-design approaches in fostering both effective technology use and social engagement, while also examining socio-ethical concerns such as cultural bias and accessibility. This review underscores the potential of GAI to support dynamic and personalized interactions, but calls for greater attention to equitable design practices and inclusive evaluation strategies.
Authors: Leigang Qu, Feng Cheng, Ziyan Yang, Qi Zhao, Shanchuan Lin, Yichun Shi, Yicong Li, Wenjie Wang, Tat-Seng Chua, Lu Jiang
Abstract: In-context image editing aims to modify images based on a contextual sequence comprising text and previously generated images. Existing methods typically depend on task-specific pipelines and expert models (e.g., segmentation and inpainting) to curate training data. In this work, we explore whether an in-context image editing model can be learned directly from videos. We introduce a scalable approach to annotate videos as interleaved multimodal sequences. To effectively learn from this data, we design a block-causal diffusion transformer trained on three proxy tasks: next-image prediction, current segmentation prediction, and next-segmentation prediction. Additionally, we propose a novel multi-turn image editing benchmark to advance research in this area. Extensive experiments demonstrate that our model exhibits strong in-context image editing capabilities and achieves state-of-the-art results on two multi-turn image editing benchmarks. Despite being trained exclusively on videos, our model also shows promising abilities in multi-concept composition, story generation, and chain-of-editing applications.
Authors: Evelyn Ma, Duo Zhou, Peizhi Niu, Huiting Zhou, Huan Zhang, Olgica Milenkovic, S. Rasoul Etesami
Abstract: Unlearning in large language models (LLMs) is becoming increasingly important due to regulatory compliance, copyright protection, and privacy concerns. However, a key challenge in LLM unlearning is unintended forgetting, where the removal of specific data inadvertently impairs the utility of the model and its retention of valuable, desired information. While prior work has primarily focused on architectural innovations, the influence of data-level factors on unlearning performance remains underexplored. As a result, existing methods often suffer from degraded retention when forgetting high-impact data. To address this, we propose GUARD-a novel framework for Guided Unlearning And Retention via Data attribution. At its core, GUARD introduces a lightweight proxy data attribution metric tailored for LLM unlearning, which quantifies the "alignment" between the forget and retain sets while remaining computationally efficient. Building on this, we design a novel unlearning objective that assigns adaptive, nonuniform unlearning weights to samples, inversely proportional to their proxy attribution scores. Through such a reallocation of unlearning power, GUARD mitigates unintended losses in retention. We provide rigorous theoretical guarantees that GUARD significantly enhances retention while maintaining forgetting metrics comparable to prior methods. Extensive experiments on the TOFU benchmark across multiple LLM architectures demonstrate that GUARD substantially improves utility preservation while ensuring effective unlearning. Notably, GUARD reduces utility sacrifice on the Retain Set by up to 194.92% in terms of Truth Ratio when forgetting 10% of the training data.
Authors: Chen Yueh-Han, Nitish Joshi, Yulin Chen, Maksym Andriushchenko, Rico Angell, He He
Abstract: Current LLM safety defenses fail under decomposition attacks, where a malicious goal is decomposed into benign subtasks that circumvent refusals. The challenge lies in the existing shallow safety alignment techniques: they only detect harm in the immediate prompt and do not reason about long-range intent, leaving them blind to malicious intent that emerges over a sequence of seemingly benign instructions. We therefore propose adding an external monitor that observes the conversation at a higher granularity. To facilitate our study of monitoring decomposition attacks, we curate the largest and most diverse dataset to date, including question-answering, text-to-image, and agentic tasks. We verify our datasets by testing them on frontier LLMs and show an 87% attack success rate on average on GPT-4o. This confirms that decomposition attack is broadly effective. Additionally, we find that random tasks can be injected into the decomposed subtasks to further obfuscate malicious intents. To defend in real time, we propose a lightweight sequential monitoring framework that cumulatively evaluates each subtask. We show that a carefully prompt engineered lightweight monitor achieves a 93% defense success rate, beating reasoning models like o3 mini as a monitor. Moreover, it remains robust against random task injection and cuts cost by 90% and latency by 50%. Our findings suggest that lightweight sequential monitors are highly effective in mitigating decomposition attacks and are viable in deployment.
Authors: Mozhi Zhang, Howe Tissue, Lu Wang, Xipeng Qiu
Abstract: We introduce~\textsc{Domain2Vec}, a novel approach that decomposes any dataset into a linear combination of several \emph{meta-domains}, a new concept designed to capture the key underlying features of datasets. \textsc{Domain2Vec} maintains a vocabulary of meta-domains and uses a classifier to decompose any given dataset into a domain vector that corresponds to a distribution over this vocabulary. These domain vectors enable the identification of the optimal data mixture for language model (LM) pretraining in a training-free manner under the \emph{\textbf{D}istribution \textbf{A}lignment \textbf{A}ssumption} (DA$^{2}$), which suggests that when the data distributions of the training set and the validation set are better aligned, a lower validation loss is achieved. Moreover, \textsc{Domain2vec} can be seamlessly integrated into previous works to model the relationship between domain vectors and LM performance, greatly enhancing the efficiency and scalability of previous methods. Extensive experiments demonstrate that \textsc{Domain2Vec} helps find the data mixture that enhances downstream task performance with minimal computational overhead. Specifically, \textsc{Domain2Vec} achieves the same validation loss on Pile-CC using only $51.5\%$ of the computation required when training on the original mixture of The Pile dataset. Under equivalent compute budget, \textsc{Domain2Vec} improves downstream performance by an average of $2.83\%$.
Authors: Lianghong Guo, Yanlin Wang, Caihua Li, Pengyu Yang, Jiachi Chen, Wei Tao, Yingtian Zou, Duyu Tang, Zibin Zheng
Abstract: Constructing large-scale datasets for the GitHub issue resolution task is crucial for both training and evaluating the software engineering capabilities of Large Language Models (LLMs). However, the traditional process for creating such benchmarks is notoriously challenging and labor-intensive, particularly in the stages of setting up evaluation environments, grading test outcomes, and validating task instances. In this paper, we propose SWE-Factory, an automated pipeline designed to address these challenges. To tackle these issues, our pipeline integrates three core automated components. First, we introduce SWE-Builder, a multi-agent system that automates evaluation environment construction, which employs four specialized agents that work in a collaborative, iterative loop and leverages an environment memory pool to enhance efficiency. Second, we introduce a standardized, exit-code-based grading method that eliminates the need for manually writing custom parsers. Finally, we automate the fail2pass validation process using these reliable exit code signals. Experiments on 671 issues across four programming languages show that our pipeline can effectively construct valid task instances; for example, with GPT-4.1-mini, our SWE-Builder constructs 269 valid instances at $0.045 per instance, while with Gemini-2.5-flash, it achieves comparable performance at the lowest cost of $0.024 per instance. We also demonstrate that our exit-code-based grading achieves 100% accuracy compared to manual inspection, and our automated fail2pass validation reaches a precision of 0.92 and a recall of 1.00. We hope our automated pipeline will accelerate the collection of large-scale, high-quality GitHub issue resolution datasets for both training and evaluation. Our code and datasets are released at https://github.com/DeepSoftwareAnalytics/swe-factory.
Authors: Aayush Karan, Kulin Shah, Sitan Chen
Abstract: There has been a flurry of activity around using pretrained diffusion models as informed data priors for solving inverse problems, and more generally around steering these models using reward models. Training-free methods like diffusion posterior sampling (DPS) and its many variants have offered flexible heuristic algorithms for these tasks, but when the reward is not informative enough, e.g., in hard inverse problems with low signal-to-noise ratio, these techniques veer off the data manifold, failing to produce realistic outputs. In this work, we devise a simple wrapper, ReGuidance, for boosting both the sample realism and reward achieved by these methods. Given a candidate solution $\hat{x}$ produced by an algorithm of the user's choice, we propose inverting the solution by running the unconditional probability flow ODE in reverse starting from $\hat{x}$, and then using the resulting latent as an initialization for DPS. We evaluate our wrapper on hard inverse problems like large box in-painting and super-resolution with high upscaling. Whereas state-of-the-art baselines visibly fail, we find that applying our wrapper on top of these baselines significantly boosts sample quality and measurement consistency. We complement these findings with theory proving that on certain multimodal data distributions, ReGuidance simultaneously boosts the reward and brings the candidate solution closer to the data manifold. To our knowledge, this constitutes the first rigorous algorithmic guarantee for DPS.
Authors: Zhaiming Shen, Alexander Hsu, Rongjie Lai, Wenjing Liao
Abstract: While in-context learning (ICL) has achieved remarkable success in natural language and vision domains, its theoretical understanding--particularly in the context of structured geometric data--remains unexplored. In this work, we initiate a theoretical study of ICL for regression of H\"older functions on manifolds. By establishing a novel connection between the attention mechanism and classical kernel methods, we derive generalization error bounds in terms of the prompt length and the number of training tasks. When a sufficient number of training tasks are observed, transformers give rise to the minimax regression rate of H\"older functions on manifolds, which scales exponentially with the intrinsic dimension of the manifold, rather than the ambient space dimension. Our result also characterizes how the generalization error scales with the number of training tasks, shedding light on the complexity of transformers as in-context algorithm learners. Our findings provide foundational insights into the role of geometry in ICL and novels tools to study ICL of nonlinear models.
Authors: Kangwei Liu, Siyuan Cheng, Bozhong Tian, Xiaozhuan Liang, Yuyang Yin, Meng Han, Ningyu Zhang, Bryan Hooi, Xi Chen, Shumin Deng
Abstract: Large language models (LLMs) have been increasingly applied to automated harmful content detection tasks, assisting moderators in identifying policy violations and improving the overall efficiency and accuracy of content review. However, existing resources for harmful content detection are predominantly focused on English, with Chinese datasets remaining scarce and often limited in scope. We present a comprehensive, professionally annotated benchmark for Chinese content harm detection, which covers six representative categories and is constructed entirely from real-world data. Our annotation process further yields a knowledge rule base that provides explicit expert knowledge to assist LLMs in Chinese harmful content detection. In addition, we propose a knowledge-augmented baseline that integrates both human-annotated knowledge rules and implicit knowledge from large language models, enabling smaller models to achieve performance comparable to state-of-the-art LLMs. Code and data are available at https://github.com/zjunlp/ChineseHarm-bench.
Authors: Yuanhui Huang, Weiliang Chen, Wenzhao Zheng, Yueqi Duan, Jie Zhou, Jiwen Lu
Abstract: Autoregressive visual generation has garnered increasing attention due to its scalability and compatibility with other modalities compared with diffusion models. Most existing methods construct visual sequences as spatial patches for autoregressive generation. However, image patches are inherently parallel, contradicting the causal nature of autoregressive modeling. To address this, we propose a Spectral AutoRegressive (SpectralAR) visual generation framework, which realizes causality for visual sequences from the spectral perspective. Specifically, we first transform an image into ordered spectral tokens with Nested Spectral Tokenization, representing lower to higher frequency components. We then perform autoregressive generation in a coarse-to-fine manner with the sequences of spectral tokens. By considering different levels of detail in images, our SpectralAR achieves both sequence causality and token efficiency without bells and whistles. We conduct extensive experiments on ImageNet-1K for image reconstruction and autoregressive generation, and SpectralAR achieves 3.02 gFID with only 64 tokens and 310M parameters. Project page: https://huang-yh.github.io/spectralar/.
Authors: Qizhe Zhang, Mengzhen Liu, Lichen Li, Ming Lu, Yuan Zhang, Junwen Pan, Qi She, Shanghang Zhang
Abstract: In multimodal large language models (MLLMs), the length of input visual tokens is often significantly greater than that of their textual counterparts, leading to a high inference cost. Many works aim to address this issue by removing redundant visual tokens. However, current approaches either rely on attention-based pruning, which retains numerous duplicate tokens, or use similarity-based pruning, overlooking the instruction relevance, consequently causing suboptimal performance. In this paper, we go beyond attention or similarity by proposing a novel visual token pruning method named CDPruner, which maximizes the conditional diversity of retained tokens. We first define the conditional similarity between visual tokens conditioned on the instruction, and then reformulate the token pruning problem with determinantal point process (DPP) to maximize the conditional diversity of the selected subset. The proposed CDPruner is training-free and model-agnostic, allowing easy application to various MLLMs. Extensive experiments across diverse MLLMs show that CDPruner establishes new state-of-the-art on various vision-language benchmarks. By maximizing conditional diversity through DPP, the selected subset better represents the input images while closely adhering to user instructions, thereby preserving strong performance even with high reduction ratios. When applied to LLaVA, CDPruner reduces FLOPs by 95\% and CUDA latency by 78\%, while maintaining 94\% of the original accuracy. Our code is available at https://github.com/Theia-4869/CDPruner.
Authors: Houyi Li, Wenzhen Zheng, Qiufeng Wang, Zhenyu Ding, Haoying Wang, Zili Wang, Shijie Xuyang, Ning Ding, Shuigeng Zhou, Xiangyu Zhang, Daxin Jiang
Abstract: Training Large Language Models (LLMs) is prohibitively expensive, creating a critical scaling gap where insights from small-scale experiments often fail to transfer to resource-intensive production systems, thereby hindering efficient innovation. To bridge this, we introduce Farseer, a novel and refined scaling law offering enhanced predictive accuracy across scales. By systematically constructing a model loss surface $L(N,D)$, Farseer achieves a significantly better fit to empirical data than prior laws (e.g., Chinchilla's law). Our methodology yields accurate, robust, and highly generalizable predictions, demonstrating excellent extrapolation capabilities, improving upon Chinchilla's law by reducing extrapolation error by 433\%. This allows for the reliable evaluation of competing training strategies across all $(N,D)$ settings, enabling conclusions from small-scale ablation studies to be confidently extrapolated to predict large-scale performance. Furthermore, Farseer provides new insights into optimal compute allocation, better reflecting the nuanced demands of modern LLM training. To validate our approach, we trained an extensive suite of approximately 1,000 LLMs across diverse scales and configurations, consuming roughly 3 million NVIDIA H100 GPU hours. We are comprehensively open-sourcing all models, data, results, and logs at https://github.com/Farseer-Scaling-Law/Farseer to foster further research.
Authors: Julius Berner, Miguel Liu-Schiaffini, Jean Kossaifi, Valentin Duruisseaux, Boris Bonev, Kamyar Azizzadenesheli, Anima Anandkumar
Abstract: A wide range of scientific problems, such as those described by continuous-time dynamical systems and partial differential equations (PDEs), are naturally formulated on function spaces. While function spaces are typically infinite-dimensional, deep learning has predominantly advanced through applications in computer vision and natural language processing that focus on mappings between finite-dimensional spaces. Such fundamental disparities in the nature of the data have limited neural networks from achieving a comparable level of success in scientific applications as seen in other fields. Neural operators are a principled way to generalize neural networks to mappings between function spaces, offering a pathway to replicate deep learning's transformative impact on scientific problems. For instance, neural operators can learn solution operators for entire classes of PDEs, e.g., physical systems with different boundary conditions, coefficient functions, and geometries. A key factor in deep learning's success has been the careful engineering of neural architectures through extensive empirical testing. Translating these neural architectures into neural operators allows operator learning to enjoy these same empirical optimizations. However, prior neural operator architectures have often been introduced as standalone models, not directly derived as extensions of existing neural network architectures. In this paper, we identify and distill the key principles for constructing practical implementations of mappings between infinite-dimensional function spaces. Using these principles, we propose a recipe for converting several popular neural architectures into neural operators with minimal modifications. This paper aims to guide practitioners through this process and details the steps to make neural operators work in practice. Our code can be found at https://github.com/neuraloperator/NNs-to-NOs
Authors: Yixin Ou, Yujie Luo, Jingsheng Zheng, Lanning Wei, Shuofei Qiao, Jintian Zhang, Da Zheng, Huajun Chen, Ningyu Zhang
Abstract: Large Language Model (LLM) agents have shown great potential in addressing real-world data science problems. LLM-driven data science agents promise to automate the entire machine learning pipeline, yet their real-world effectiveness remains limited. Existing frameworks depend on rigid, pre-defined workflows and inflexible coding strategies; consequently, they excel only on relatively simple, classical problems and fail to capture the empirical expertise that human practitioners bring to complex, innovative tasks. In this work, we introduce AutoMind, an adaptive, knowledgeable LLM-agent framework that overcomes these deficiencies through three key advances: (1) a curated expert knowledge base that grounds the agent in domain expert knowledge, (2) an agentic knowledgeable tree search algorithm that strategically explores possible solutions, and (3) a self-adaptive coding strategy that dynamically tailors code generation to task complexity. Evaluations on two automated data science benchmarks demonstrate that AutoMind delivers superior performance versus state-of-the-art baselines. Additional analyses confirm favorable effectiveness, efficiency, and qualitative solution quality, highlighting AutoMind as an efficient and robust step toward fully automated data science.
Authors: Donghoon Ahn, Jiwon Kang, Sanghyun Lee, Minjae Kim, Jaewon Min, Wooseok Jang, Saungwu Lee, Sayak Paul, Susung Hong, Seungryong Kim
Abstract: Recent guidance methods in diffusion models steer reverse sampling by perturbing the model to construct an implicit weak model and guide generation away from it. Among these approaches, attention perturbation has demonstrated strong empirical performance in unconditional scenarios where classifier-free guidance is not applicable. However, existing attention perturbation methods lack principled approaches for determining where perturbations should be applied, particularly in Diffusion Transformer (DiT) architectures where quality-relevant computations are distributed across layers. In this paper, we investigate the granularity of attention perturbations, ranging from the layer level down to individual attention heads, and discover that specific heads govern distinct visual concepts such as structure, style, and texture quality. Building on this insight, we propose "HeadHunter", a systematic framework for iteratively selecting attention heads that align with user-centric objectives, enabling fine-grained control over generation quality and visual attributes. In addition, we introduce SoftPAG, which linearly interpolates each selected head's attention map toward an identity matrix, providing a continuous knob to tune perturbation strength and suppress artifacts. Our approach not only mitigates the oversmoothing issues of existing layer-level perturbation but also enables targeted manipulation of specific visual styles through compositional head selection. We validate our method on modern large-scale DiT-based text-to-image models including Stable Diffusion 3 and FLUX.1, demonstrating superior performance in both general quality enhancement and style-specific guidance. Our work provides the first head-level analysis of attention perturbation in diffusion models, uncovering interpretable specialization within attention layers and enabling practical design of effective perturbation strategies.
Authors: Sebastian Sanokowski, Lukas Gruber, Christoph Bartmann, Sepp Hochreiter, Sebastian Lehner
Abstract: Diffusion bridges are a promising class of deep-learning methods for sampling from unnormalized distributions. Recent works show that the Log Variance (LV) loss consistently outperforms the reverse Kullback-Leibler (rKL) loss when using the reparametrization trick to compute rKL-gradients. While the on-policy LV loss yields identical gradients to the rKL loss when combined with the log-derivative trick for diffusion samplers with non-learnable forward processes, this equivalence does not hold for diffusion bridges or when diffusion coefficients are learned. Based on this insight we argue that for diffusion bridges the LV loss does not represent an optimization objective that can be motivated like the rKL loss via the data processing inequality. Our analysis shows that employing the rKL loss with the log-derivative trick (rKL-LD) does not only avoid these conceptual problems but also consistently outperforms the LV loss. Experimental results with different types of diffusion bridges on challenging benchmarks show that samplers trained with the rKL-LD loss achieve better performance. From a practical perspective we find that rKL-LD requires significantly less hyperparameter optimization and yields more stable training behavior.
Authors: Yushan Zhu, Wen Zhang, Zhiqiang Liu, Mingyang Chen, Lei Liang, Huajun Chen
Abstract: Knowledge Graph Embedding (KGE) is a common approach for Knowledge Graphs (KGs) in AI tasks. Embedding dimensions depend on application scenarios. Requiring a new dimension means training a new KGE model from scratch, increasing cost and limiting efficiency and flexibility. In this work, we propose a novel KGE training framework MED. It allows one training to obtain a croppable KGE model for multiple scenarios with different dimensional needs. Sub-models of required dimensions can be directly cropped and used without extra training. In MED, we propose a mutual learning mechanism to improve the low-dimensional sub-models and make high-dimensional sub-models retain the low-dimensional sub-models' capacity, an evolutionary improvement mechanism to promote the high-dimensional sub-models to master the triple that the low-dimensional sub-models can not, and a dynamic loss weight to adaptively balance the multiple losses. Experiments on 4 KGE models across 4 standard KG completion datasets, 3 real-world scenarios using a large-scale KG, and extending MED to the BERT language model demonstrate its effectiveness, high efficiency, and flexible extensibility.
Authors: Xingrui Gu, Guanren Qiao, Chuyi Jiang
Abstract: Traditional reinforcement learning (RL) methods mainly rely on trial-and-error exploration, often lacking mechanisms to guide agents toward more informative decision-making and struggling to leverage past experiences, resulting in low sample efficiency. To overcome this issue, we propose an innovative framework inspired by cognitive principles: Cognitive Belief-Driven Reinforcement Learning (CBD-RL). By incorporating cognitive heuristics, CBD-RL transforms conventional trial-and-error learning into a more structured and guided learning paradigm, simulating the human reasoning process. This framework's core is a belief system that optimizes action probabilities by integrating feedback with prior experience, thus enhancing decision making under uncertainty. It also organizes state-action pairs into meaningful categories, promoting generalization and improving sample efficiency. The concrete implementations of this framework, CBDQ, CBDPPO, and CBDSAC, demonstrate superior performance in discrete and continuous action spaces in diverse environments such as Atari and MuJoCo. By bridging cognitive science and reinforcement learning, this research opens a new avenue for developing RL systems that are more interpretable, efficient, and cognitively inspired.
Authors: Weihua Du, Qiushi Lyu, Jiaming Shan, Zhenting Qi, Hongxin Zhang, Sunli Chen, Andi Peng, Tianmin Shu, Kwonjoon Lee, Behzad Dariush, Chuang Gan
Abstract: We introduce Constrained Human-AI Cooperation (CHAIC), an inclusive embodied social intelligence challenge designed to test social perception and cooperation in embodied agents. In CHAIC, the goal is for an embodied agent equipped with egocentric observations to assist a human who may be operating under physical constraints -- e.g., unable to reach high places or confined to a wheelchair -- in performing common household or outdoor tasks as efficiently as possible. To achieve this, a successful helper must: (1) infer the human's intents and constraints by following the human and observing their behaviors (social perception), and (2) make a cooperative plan tailored to the human partner to solve the task as quickly as possible, working together as a team (cooperative planning). To benchmark this challenge, we create four new agents with real physical constraints and eight long-horizon tasks featuring both indoor and outdoor scenes with various constraints, emergency events, and potential risks. We benchmark planning- and learning-based baselines on the challenge and introduce a new method that leverages large language models and behavior modeling. Empirical evaluations demonstrate the effectiveness of our benchmark in enabling systematic assessment of key aspects of machine social intelligence. Our benchmark and code are publicly available at https://github.com/UMass-Embodied-AGI/CHAIC.
Authors: Junyang Cai, Serdar Kadioglu, Bistra Dilkina
Abstract: Mixed-integer programming (MIP) is a powerful paradigm for modeling and solving various important combinatorial optimization problems. Recently, learning-based approaches have shown a potential to speed up MIP solving via offline training that then guides important design decisions during the search. However, a significant drawback of these methods is their heavy reliance on offline training, which requires collecting training datasets and computationally costly training epochs yet offering only limited generalization to unseen (larger) instances. In this paper, we propose Balans, an adaptive meta-solver for MIPs with online learning capability that does not require any supervision or apriori training. At its core, Balans is based on adaptive large-neighborhood search, operating on top of an MIP solver by successive applications of destroy and repair neighborhood operators. During the search, the selection among different neighborhood definitions is guided on the fly for the instance at hand via multi-armed bandit algorithms. Our extensive experiments on hard optimization instances show that Balans offers significant performance gains over the default MIP solver, is better than committing to any single best neighborhood, and improves over the state-of-the-art large-neighborhood search for MIPs. Finally, we release Balans as a highly configurable, MIP solver agnostic, open-source software.
Authors: Junyang Cai, Taoan Huang, Bistra Dilkina
Abstract: Mixed Integer Linear Programs (MILPs) are highly flexible and powerful tools for modeling and solving complex real-world combinatorial optimization problems. Recently, machine learning (ML)-guided approaches have demonstrated significant potential in improving MILP-solving efficiency. However, these methods typically rely on separate offline data collection and training processes, which limits their scalability and adaptability. This paper introduces the first multi-task learning framework for ML-guided MILP solving. The proposed framework provides MILP embeddings helpful in guiding MILP solving across solvers (e.g., Gurobi and SCIP) and across tasks (e.g., Branching and Solver configuration). Through extensive experiments on three widely used MILP benchmarks, we demonstrate that our multi-task learning model performs similarly to specialized models within the same distribution. Moreover, it significantly outperforms them in generalization across problem sizes and tasks.
Authors: Matthew Riemer, Zahra Ashktorab, Djallel Bouneffouf, Payel Das, Miao Liu, Justin D. Weisz, Murray Campbell
Abstract: Our paper argues that the majority of theory of mind benchmarks are broken because of their inability to directly test how large language models (LLMs) adapt to new partners. This problem stems from the fact that theory of mind benchmarks for LLMs are overwhelmingly inspired by the methods used to test theory of mind in humans and fall victim to a fallacy of attributing human-like qualities to AI agents. We expect that humans will engage in a consistent reasoning process across various questions about a situation, but this is known to not be the case for current LLMs. Most theory of mind benchmarks only measure what we call literal theory of mind: the ability to predict the behavior of others. However, this type of metric is only informative when agents exhibit self-consistent reasoning. Thus, we introduce the concept of functional theory of mind: the ability to adapt to agents in-context following a rational response to their behavior. We find that many open source LLMs are capable of displaying strong literal theory of mind capabilities, but seem to struggle with functional theory of mind -- even with exceedingly simple partner policies. Simply put, strong literal theory of mind performance does not necessarily imply strong functional theory of mind performance or vice versa. Achieving functional theory of mind, particularly over long interaction horizons with a partner, is a significant challenge deserving a prominent role in any meaningful LLM theory of mind evaluation.
Authors: Shirley Wu, Michel Galley, Baolin Peng, Hao Cheng, Gavin Li, Yao Dou, Weixin Cai, James Zou, Jure Leskovec, Jianfeng Gao
Abstract: Large Language Models are typically trained with next-turn rewards, limiting their ability to optimize for long-term interaction. As a result, they often respond passively to ambiguous or open-ended user requests, failing to help users reach their ultimate intents and leading to inefficient conversations. To address these limitations, we introduce CollabLLM, a novel and general training framework that enhances multiturn human-LLM collaboration. Its key innovation is a collaborative simulation that estimates the long-term contribution of responses using Multiturn-aware Rewards. By reinforcement fine-tuning these rewards, CollabLLM goes beyond responding to user requests, and actively uncovers user intent and offers insightful suggestions-a key step towards more human-centered AI. We also devise a multiturn interaction benchmark with three challenging tasks such as document creation. CollabLLM significantly outperforms our baselines with averages of 18.5% higher task performance and 46.3% improved interactivity by LLM judges. Finally, we conduct a large user study with 201 judges, where CollabLLM increases user satisfaction by 17.6% and reduces user spent time by 10.4%.
Authors: Mingyu Kim, Dongjun Kim, Amman Yusuf, Stefano Ermon, Mijung Park
Abstract: There is growing concern over the safety of powerful diffusion models (DMs), as they are often misused to produce inappropriate, not-safe-for-work (NSFW) content or generate copyrighted material or data of individuals who wish to be forgotten. Many existing methods tackle these issues by heavily relying on text-based negative prompts or extensively retraining DMs to eliminate certain features or samples. In this paper, we take a radically different approach, directly modifying the sampling trajectory by leveraging a negation set (e.g., unsafe images, copyrighted data, or datapoints needed to be excluded) to avoid specific regions of data distribution, without needing to retrain or fine-tune DMs. We formally derive the relationship between the expected denoised samples that are safe and those that are not safe, leading to our $\textit{safe}$ denoiser which ensures its final samples are away from the area to be negated. Inspired by the derivation, we develop a practical algorithm that successfully produces high-quality samples while avoiding negation areas of the data distribution in text-conditional, class-conditional, and unconditional image generation scenarios. These results hint at the great potential of our training-free safe denoiser for using DMs more safely.
Authors: Yuichi Inoue, Kou Misaki, Yuki Imajuku, So Kuroki, Taishi Nakamura, Takuya Akiba
Abstract: Recent advances demonstrate that increasing inference-time computation can significantly boost the reasoning capabilities of large language models (LLMs). Although repeated sampling (i.e., generating multiple candidate outputs) is a highly effective strategy, it does not leverage external feedback signals for refinement, which are often available in tasks like coding. In this work, we propose Adaptive Branching Monte Carlo Tree Search (AB-MCTS), a novel inference-time framework that generalizes repeated sampling with principled multi-turn exploration and exploitation. At each node in the search tree, AB-MCTS dynamically decides whether to "go wider" by expanding new candidate responses or "go deeper" by revisiting existing ones based on external feedback signals. We evaluate our method on complex coding and engineering tasks using frontier models. Empirical results show that AB-MCTS consistently outperforms both repeated sampling and standard MCTS, underscoring the importance of combining the response diversity of LLMs with multi-turn solution refinement for effective inference-time scaling.
Authors: Roie Kazoom, Raz Lapid, Moshe Sipper, Ofer Hadar
Abstract: Adversarial patch attacks pose a major threat to vision systems by embedding localized perturbations that mislead deep models. Traditional defense methods often require retraining or fine-tuning, making them impractical for real-world deployment. We propose a training-free Visual Retrieval-Augmented Generation (VRAG) framework that integrates Vision-Language Models (VLMs) for adversarial patch detection. By retrieving visually similar patches and images that resemble stored attacks in a continuously expanding database, VRAG performs generative reasoning to identify diverse attack types, all without additional training or fine-tuning. We extensively evaluate open-source large-scale VLMs, including Qwen-VL-Plus, Qwen2.5-VL-72B, and UI-TARS-72B-DPO, alongside Gemini-2.0, a closed-source model. Notably, the open-source UI-TARS-72B-DPO model achieves up to 95 percent classification accuracy, setting a new state-of-the-art for open-source adversarial patch detection. Gemini-2.0 attains the highest overall accuracy, 98 percent, but remains closed-source. Experimental results demonstrate VRAG's effectiveness in identifying a variety of adversarial patches with minimal human annotation, paving the way for robust, practical defenses against evolving adversarial patch attacks.
Authors: Cassidy Laidlaw, Eli Bronstein, Timothy Guo, Dylan Feng, Lukas Berglund, Justin Svegliato, Stuart Russell, Anca Dragan
Abstract: Assistance games are a promising alternative to reinforcement learning from human feedback (RLHF) for training AI assistants. Assistance games resolve key drawbacks of RLHF, such as incentives for deceptive behavior, by explicitly modeling the interaction between assistant and user as a two-player game where the assistant cannot observe their shared goal. Despite their potential, assistance games have only been explored in simple settings. Scaling them to more complex environments is difficult because it requires both solving intractable decision-making problems under uncertainty and accurately modeling human users' behavior. We present the first scalable approach to solving assistance games and apply it to a new, challenging Minecraft-based assistance game with over $10^{400}$ possible goals. Our approach, AssistanceZero, extends AlphaZero with a neural network that predicts human actions and rewards, enabling it to plan under uncertainty. We show that AssistanceZero outperforms model-free RL algorithms and imitation learning in the Minecraft-based assistance game. In a human study, our AssistanceZero-trained assistant significantly reduces the number of actions participants take to complete building tasks in Minecraft. Our results suggest that assistance games are a tractable framework for training effective AI assistants in complex environments. Our code and models are available at https://github.com/cassidylaidlaw/minecraft-building-assistance-game.
URLs: https://github.com/cassidylaidlaw/minecraft-building-assistance-game.
Authors: Chengwei Liu, Chong Wang, Jiayue Cao, Jingquan Ge, Kun Wang, Lyuye Zhang, Ming-Ming Cheng, Penghai Zhao, Tianlin Li, Xiaojun Jia, Xiang Li, Xinfeng Li, Yang Liu, Yebo Feng, Yihao Huang, Yijia Xu, Yuqiang Sun, Zhenhong Zhou, Zhengzi Xu
Abstract: This paper introduces Agent-Based Auto Research, a structured multi-agent framework designed to automate, coordinate, and optimize the full lifecycle of scientific research. Leveraging the capabilities of large language models (LLMs) and modular agent collaboration, the system spans all major research phases, including literature review, ideation, methodology planning, experimentation, paper writing, peer review response, and dissemination. By addressing issues such as fragmented workflows, uneven methodological expertise, and cognitive overload, the framework offers a systematic and scalable approach to scientific inquiry. Preliminary explorations demonstrate the feasibility and potential of Auto Research as a promising paradigm for self-improving, AI-driven research processes.
Authors: Nathalie Sanghikian, Rafael Meirelles, Rafael Martinelli, Anand Subramanian
Abstract: Maritime Inventory Routing Problem (MIRP) plays a crucial role in the integration of global maritime commerce levels. However, there are still no well-established methodologies capable of efficiently solving large MIRP instances or their variants due to the high complexity of the problem. The adoption of exact methods, typically based on Mixed Integer Programming (MIP), for daily operations is nearly impractical due to the CPU time required, as planning must be executed multiple times while ensuring high-quality results within acceptable time limits. Non-MIP-based heuristics are less frequently applied due to the highly constrained nature of the problem, which makes even the construction of an effective initial solution challenging. Papageorgiou et al. (2014) introduced a single-product MIRP as the foundation for MIRPLib, aiming to provide a collection of publicly available benchmark instances. However, only a few studies that propose new methodologies have been published since then. To encourage the use of MIRPLib and facilitate result comparisons, this study presents a heuristic approach that does not rely on mathematical optimization techniques to solve a deterministic, finite-horizon, single-product MIRP. The proposed heuristic combines a variation of a Beam Search algorithm with an Iterated Local Search procedure. Among the 72 instances tested, the developed methodology can improve the best-known solution for 19 instances within an acceptable CPU time.
Authors: Shenzhe Zhu, Jiao Sun, Yi Nian, Tobin South, Alex Pentland, Jiaxin Pei
Abstract: AI agents are increasingly used in consumer-facing applications to assist with tasks such as product search, negotiation, and transaction execution. In this paper, we explore a future scenario where both consumers and merchants authorize AI agents to fully automate negotiations and transactions. We aim to answer two key questions: (1) Do different LLM agents vary in their ability to secure favorable deals for users? (2) What risks arise from fully automating deal-making with AI agents in consumer markets? To address these questions, we develop an experimental framework that evaluates the performance of various LLM agents in real-world negotiation and transaction settings. Our findings reveal that AI-mediated deal-making is an inherently imbalanced game -- different agents achieve significantly different outcomes for their users. Moreover, behavioral anomalies in LLMs can result in financial losses for both consumers and merchants, such as overspending or accepting unreasonable deals. These results underscore that while automation can improve efficiency, it also introduces substantial risks. Users should exercise caution when delegating business decisions to AI agents.
Authors: Ruonan Wang, Runxi Wang, Yunwen Shen, Chengfeng Wu, Qinglin Zhou, Rohitash Chandra
Abstract: Large Language Models (LLMs) have shown impressive performance on a range of educational tasks, but are still understudied for their potential to solve mathematical problems. In this study, we compare three prominent LLMs, including GPT-4o, DeepSeek-V3, and Gemini-2.0, on three mathematics datasets of varying complexities (GSM8K, MATH500, and UNSW datasets). We take a five-dimensional approach based on the Structured Chain-of-Thought (SCoT) framework to assess final answer correctness, step completeness, step validity, intermediate calculation accuracy, and problem comprehension. The results show that GPT-4o is the most stable and consistent in performance across all the datasets, but particularly it performs outstandingly in high-level questions of the UNSW dataset. DeepSeek-V3 is competitively strong in well-structured domains such as optimisation, but suffers from fluctuations in accuracy in statistical inference tasks. Gemini-2.0 shows strong linguistic understanding and clarity in well-structured problems but performs poorly in multi-step reasoning and symbolic logic. Our error analysis reveals particular deficits in each model: GPT-4o is at times lacking in sufficient explanation or precision; DeepSeek-V3 leaves out intermediate steps; and Gemini-2.0 is less flexible in mathematical reasoning in higher dimensions.
Authors: Yu Ma, Xiao Li, Chongtao Guo, Le Liang, Shi Jin
Abstract: This paper investigates a joint phase design and resource allocation problem in downlink reconfigurable intelligent surface (RIS)-assisted orthogonal frequency division multiplexing (OFDM) systems to optimize average delay, where data packets for each user arrive at the base station stochastically. The sequential optimization problem is inherently a Markov decision process (MDP), making it fall within the scope of reinforcement learning. To effectively handle the mixed action space and reduce the state space dimensionality, a hybrid deep reinforcement learning (DRL) approach is proposed. Specifically, proximal policy optimization (PPO)-$\Theta$ is employed to optimize RIS phase shift design, while PPO-N is responsible for subcarrier allocation decisions. To further mitigate the curse of dimensionality associated with subcarrier allocation, a multi-agent strategy is introduced to optimize subcarrier allocation indicater more efficiently. Moreover, to achieve more adaptive resource allocation and accurately capture network dynamics, key factors closely related to average delay, including the number of backlogged packets in buffers and the current packet arrivals, are incorporated into the state space. Furthermore, a transfer learning framework is introduced to enhance training efficiency and accelerate convergence. Simulation results demonstrate that the proposed algorithm significantly reduces average delay, enhances resource allocation efficiency, and achieves superior system robustness and fairness compared to baseline methods.
Authors: Li Liu, Heng Yong
Abstract: Recently, machine learning methods have gained significant traction in scientific computing, particularly for solving Partial Differential Equations (PDEs). However, methods based on deep neural networks (DNNs) often lack convergence guarantees and computational efficiency compared to traditional numerical schemes. This work introduces DeePoly, a novel framework that transforms the solution paradigm from pure non-convex parameter optimization to a two-stage approach: first employing a DNN to capture complex global features, followed by linear space optimization with combined DNN-extracted features (Scoper) and polynomial basis functions (Sniper). This strategic combination leverages the complementary strengths of both methods -- DNNs excel at approximating complex global features (i.e., high-gradient features) and stabilize the polynomial approximation while polynomial bases provide high-precision local corrections with convergence guarantees. Theoretical analysis and numerical experiments demonstrate that this approach significantly enhances both high-order accuracy and efficiency across diverse problem types while maintaining mesh-free and scheme-free properties. This paper also serves as a theoretical exposition for the open-source project DeePoly.
Authors: Lucas Irwin, Arda Kaz, Peiyao Sheng, Sewoong Oh, Pramod Viswanath
Abstract: Law has long been a domain that has been popular in natural language processing (NLP) applications. Reasoning (ratiocination and the ability to make connections to precedent) is a core part of the practice of the law in the real world. Nevertheless, while multiple legal datasets exist, none have thus far focused specifically on reasoning tasks. We focus on a specific aspect of the legal landscape by introducing a corporate governance reasoning benchmark (CHANCERY) to test a model's ability to reason about whether executive/board/shareholder's proposed actions are consistent with corporate governance charters. This benchmark introduces a first-of-its-kind corporate governance reasoning test for language models - modeled after real world corporate governance law. The benchmark consists of a corporate charter (a set of governing covenants) and a proposal for executive action. The model's task is one of binary classification: reason about whether the action is consistent with the rules contained within the charter. We create the benchmark following established principles of corporate governance - 24 concrete corporate governance principles established in and 79 real life corporate charters selected to represent diverse industries from a total dataset of 10k real life corporate charters. Evaluations on state-of-the-art (SOTA) reasoning models confirm the difficulty of the benchmark, with models such as Claude 3.7 Sonnet and GPT-4o achieving 64.5% and 75.2% accuracy respectively. Reasoning agents exhibit superior performance, with agents based on the ReAct and CodeAct frameworks scoring 76.1% and 78.1% respectively, further confirming the advanced legal reasoning capabilities required to score highly on the benchmark. We also conduct an analysis of the types of questions which current reasoning models struggle on, revealing insights into the legal reasoning capabilities of SOTA models.
Authors: Suhana Bedi, Iddah Mlauzi, Daniel Shin, Sanmi Koyejo, Nigam H. Shah
Abstract: Multi-agent artificial intelligence systems are increasingly deployed in clinical settings, yet the relationship between component-level optimization and system-wide performance remains poorly understood. We evaluated this relationship using 2,400 real patient cases from the MIMIC-CDM dataset across four abdominal pathologies (appendicitis, pancreatitis, cholecystitis, diverticulitis), decomposing clinical diagnosis into information gathering, interpretation, and differential diagnosis. We evaluated single agent systems (one model performing all tasks) against multi-agent systems (specialized models for each task) using comprehensive metrics spanning diagnostic outcomes, process adherence, and cost efficiency. Our results reveal a paradox: while multi-agent systems generally outperformed single agents, the component-optimized or Best of Breed system with superior components and excellent process metrics (85.5% information accuracy) significantly underperformed in diagnostic accuracy (67.7% vs. 77.4% for a top multi-agent system). This finding underscores that successful integration of AI in healthcare requires not just component level optimization but also attention to information flow and compatibility between agents. Our findings highlight the need for end to end system validation rather than relying on component metrics alone.
Authors: John Laird, Christian Lebiere, Paul Rosenbloom, Andrea Stocco
Abstract: The Common Model of Cognition (CMC) provides an abstract characterization of the structure and processing required by a cognitive architecture for human-like minds. We propose a unified approach to integrating metacognition within the CMC. We propose that metacognition involves reasoning over explicit representations of an agent's cognitive capabilities and processes in working memory. Our proposal exploits the existing cognitive capabilities of the CMC, making minimal extensions in the structure and information available within working memory. We provide examples of metacognition within our proposal.
Authors: Jixiang Hong, Yiran Zhang, Guanzhong Wang, Yi Liu, Ji-Rong Wen, Rui Yan
Abstract: Building upon large language models (LLMs), recent large multimodal models (LMMs) unify cross-model understanding and generation into a single framework. However, LMMs still struggle to achieve accurate image-text alignment, prone to generating text responses contradicting the visual input or failing to follow the text-to-image prompts. Current solutions require external supervision (e.g., human feedback or reward models) and only address unidirectional tasks-either understanding or generation. In this work, based on the observation that understanding and generation are inverse dual tasks, we introduce a self-supervised dual reward mechanism to reinforce the understanding and generation capabilities of LMMs. Specifically, we sample multiple outputs for a given input in one task domain, then reverse the input-output pairs to compute the dual likelihood of the model as self-rewards for optimization. Extensive experimental results on visual understanding and generation benchmarks demonstrate that our method can effectively enhance the performance of the model without any external supervision, especially achieving remarkable improvements in text-to-image tasks.
Authors: Bernardo Subercaseaux, Marijn J. H. Heule
Abstract: A packing $k$-coloring is a natural variation on the standard notion of graph $k$-coloring, where vertices are assigned numbers from $\{1, \ldots, k\}$, and any two vertices assigned a common color $c \in \{1, \ldots, k\}$ need to be at a distance greater than $c$ (as opposed to $1$, in standard graph colorings). Despite a sequence of incremental work, determining the packing chromatic number of the infinite square grid has remained an open problem since its introduction in 2002. We culminate the search by proving this number to be 15. We achieve this result by improving the best-known method for this problem by roughly two orders of magnitude. The most important technique to boost performance is a novel and surprisingly effective propositional encoding. Additionally, we developed a new symmetry-breaking method. Since both new techniques are more complex than existing techniques for this problem, a verified approach is required to trust them. We include both techniques in a proof of unsatisfiability, reducing the trusted core to the correctness of the direct encoding.
Authors: Liu Ziyin, Hongchao Li, Masahito Ueda
Abstract: The stochastic gradient descent (SGD) algorithm is the algorithm we use to train neural networks. However, it remains poorly understood how the SGD navigates the highly nonlinear and degenerate loss landscape of a neural network. In this work, we show that the minibatch noise of SGD regularizes the solution towards a noise-balanced solution whenever the loss function contains a rescaling parameter symmetry. Because the difference between a simple diffusion process and SGD dynamics is the most significant when symmetries are present, our theory implies that the loss function symmetries constitute an essential probe of how SGD works. We then apply this result to derive the stationary distribution of stochastic gradient flow for a diagonal linear network with arbitrary depth and width. The stationary distribution exhibits complicated nonlinear phenomena such as phase transitions, broken ergodicity, and fluctuation inversion. These phenomena are shown to exist uniquely in deep networks, implying a fundamental difference between deep and shallow models.
Authors: Wenbin Guo, Zhao Li, Xin Wang, Zirui Chen, Jun Zhao, Jianxin Li, Ye Yuan
Abstract: Knowledge graphs often suffer from incompleteness issues, which can be alleviated through information completion. However, current state-of-the-art deep knowledge convolutional embedding models rely on external convolution kernels and conventional convolution processes, which limits the feature interaction capability of the model. This paper introduces a novel dynamic convolutional embedding model, ConvD, which directly reshapes relation embeddings into multiple internal convolution kernels. This approach effectively enhances the feature interactions between relation embeddings and entity embeddings. Simultaneously, we incorporate a priori knowledge-optimized attention mechanism that assigns different contribution weight coefficients to the multiple relation convolution kernels in dynamic convolution, further boosting the expressive power of the model. Extensive experiments on various datasets show that our proposed model consistently outperforms the state-of-the-art baseline methods, with average improvements ranging from 3.28% to 14.69% across all model evaluation metrics, while the number of parameters is reduced by 50.66% to 85.40% compared to other state-of-the-art models.
Authors: Shyam Nuggehalli, Jifan Zhang, Lalit Jain, Robert Nowak
Abstract: Class imbalance severely impacts machine learning performance on minority classes in real-world applications. While various solutions exist, active learning offers a fundamental fix by strategically collecting balanced, informative labeled examples from abundant unlabeled data. We introduce DIRECT, an algorithm that identifies class separation boundaries and selects the most uncertain nearby examples for annotation. By reducing the problem to one-dimensional active learning, DIRECT leverages established theory to handle batch labeling and label noise -- another common challenge in data annotation that particularly affects active learning methods. Our work presents the first comprehensive study of active learning under both class imbalance and label noise. Extensive experiments on imbalanced datasets show DIRECT reduces annotation costs by over 60\% compared to state-of-the-art active learning methods and over 80\% versus random sampling, while maintaining robustness to label noise.
Authors: Kahraman Kostas, Mike Just, Michael A. Lones
Abstract: Previous research on behavior-based attack detection for networks of IoT devices has resulted in machine learning models whose ability to adapt to unseen data is limited and often not demonstrated. This paper presents IoTGeM, an approach for modeling IoT network attacks that focuses on generalizability, yet also leads to better detection and performance. We first introduce an improved rolling window approach for feature extraction. To reduce overfitting, we then apply a multi-step feature selection process where a Genetic Algorithm (GA) is uniquely guided by exogenous feedback from a separate, independent dataset. To prevent common data leaks that have limited previous models, we build and test our models using strictly isolated train and test datasets. The resulting models are rigorously evaluated using a diverse portfolio of machine learning algorithms and datasets. Our window-based models demonstrate superior generalization compared to traditional flow-based models, particularly when tested on unseen datasets. On these stringent, cross-dataset tests, IoTGeM achieves F1 scores of 99\% for ACK, HTTP, SYN, MHD, and PS attacks, as well as a 94\% F1 score for UDP attacks. Finally, we build confidence in the models by using the SHAP (SHapley Additive exPlanations) explainable AI technique, allowing us to identify the specific features that underlie the accurate detection of attacks.
Authors: Longkun Guo, Chaoqi Jia, Kewen Liao, Zhigang Lu, Minhui Xue
Abstract: Center-based clustering has attracted significant research interest from both theory and practice. In many practical applications, input data often contain background knowledge that can be used to improve clustering results. In this work, we build on widely adopted $k$-center clustering and model its input background knowledge as must-link (ML) and cannot-link (CL) constraint sets. However, most clustering problems including $k$-center are inherently $\mathcal{NP}$-hard, while the more complex constrained variants are known to suffer severer approximation and computation barriers that significantly limit their applicability. By employing a suite of techniques including reverse dominating sets, linear programming (LP) integral polyhedron, and LP duality, we arrive at the first efficient approximation algorithm for constrained $k$-center with the best possible ratio of 2. We also construct competitive baseline algorithms and empirically evaluate our approximation algorithm against them on a variety of real datasets. The results validate our theoretical findings and demonstrate the great advantages of our algorithm in terms of clustering cost, clustering quality, and running time.
Authors: Linxi Zhao, Yihe Deng, Weitong Zhang, Quanquan Gu
Abstract: The advancement of Large Vision-Language Models (LVLMs) has increasingly highlighted the critical issue of their tendency to hallucinate non-existing objects in the images. To address this issue, previous works focused on using specially curated datasets or powerful LLMs to rectify the outputs of LVLMs. However, these approaches require either costly training or fine-tuning, or API access to proprietary LLMs for post-generation correction. In response to these limitations, we propose Mitigating hallucinAtion via image-gRounded guIdaNcE (MARINE), a framework that is both training-free and API-free. MARINE effectively and efficiently reduces object hallucinations during inference by introducing image-grounded guidance to LVLMs. This is achieved by leveraging open-source vision models to extract object-level information, thereby enhancing the precision of LVLM-generated content. Our framework's flexibility further allows for the integration of multiple vision models, enabling more reliable and robust object-level guidance. Through comprehensive evaluations across 5 popular LVLMs with diverse evaluation metrics and benchmarks, we demonstrate the effectiveness of MARINE, which even outperforms existing fine-tuning-based methods. Remarkably, it reduces hallucinations consistently in GPT-4V-assisted evaluation while maintaining the detailedness of LVLMs' generations. We release our code at https://github.com/Linxi-ZHAO/MARINE.
Authors: Sara Abdali, Richard Anarfi, CJ Barberan, Jia He, Erfan Shayegani
Abstract: Large language models (LLMs) have significantly transformed the landscape of Natural Language Processing (NLP). Their impact extends across a diverse spectrum of tasks, revolutionizing how we approach language understanding and generations. Nevertheless, alongside their remarkable utility, LLMs introduce critical security and risk considerations. These challenges warrant careful examination to ensure responsible deployment and safeguard against potential vulnerabilities. This research paper thoroughly investigates security and privacy concerns related to LLMs from five thematic perspectives: security and privacy concerns, vulnerabilities against adversarial attacks, potential harms caused by misuses of LLMs, mitigation strategies to address these challenges while identifying limitations of current strategies. Lastly, the paper recommends promising avenues for future research to enhance the security and risk management of LLMs.
Authors: Zhenhailong Wang, Joy Hsu, Xingyao Wang, Kuan-Hao Huang, Manling Li, Jiajun Wu, Heng Ji
Abstract: Despite significant advancements, large multimodal models (LMMs) still struggle to bridge the gap between low-level visual perception -- focusing on shapes, sizes, and layouts -- and high-level language reasoning, such as semantics and logic. This limitation is evident in tasks that require precise visual perception, like comparing geometric properties or solving visual reasoning problems. To study this failure mode, we focus on vector graphics -- images composed of 2D objects and shapes, prevalent in LMM-based tasks in web, design, and OS environments. We identify two key research questions: how can we enable precise visual perception, and how can we facilitate high-level reasoning based on such low-level perceptions? To capture fine visual details, we use Scalable Vector Graphics (SVG) for accurate encoding of visual scenes. However, SVGs are not readily interpretable by LMMs in a zero-shot manner. To tackle this, we propose the Visually Descriptive Language Model (VDLM), which introduces a Primal Visual Description (PVD) as an intermediate textual representation. PVD translates SVGs into a text-based abstraction consisting of primitive attributes (e.g., shape, position, measurement) and their corresponding values. PVD can be learned using task-agnostic synthesized data and represents visual primitives that are universal across vector graphics. This abstraction is more structured, allowing for direct interpretation by foundation models for zero-shot generalization. Without human-annotated data, empirical results show that VDLM significantly improves state-of-the-art LMMs like GPT-4o on various multimodal perception and reasoning tasks. Extensive analyses of VDLM show improved interpretability due to its disentangled perception and reasoning. We also demonstrate a positive correlation between PVD quality and task performance. Project page: https://mikewangwzhl.github.io/VDLM/
Authors: Eden Saig, Ohad Einav, Inbal Talgam-Cohen
Abstract: While the success of large language models (LLMs) increases demand for machine-generated text, current pay-per-token pricing schemes create a misalignment of incentives known in economics as moral hazard: Text-generating agents have strong incentive to cut costs by preferring a cheaper model over the cutting-edge one, and this can be done "behind the scenes" since the agent performs inference internally. In this work, we approach this issue from an economic perspective, by proposing a pay-for-performance, contract-based framework for incentivizing quality. We study a principal-agent game where the agent generates text using costly inference, and the contract determines the principal's payment for the text according to an automated quality evaluation. Since standard contract theory is inapplicable when internal inference costs are unknown, we introduce cost-robust contracts. As our main theoretical contribution, we characterize optimal cost-robust contracts through a direct correspondence to optimal composite hypothesis tests from statistics, generalizing a result of Saig et al. (NeurIPS'23). We evaluate our framework empirically by deriving contracts for a range of objectives and LLM evaluation benchmarks, and find that cost-robust contracts sacrifice only a marginal increase in objective value compared to their cost-aware counterparts.
Authors: Lucky Susanto, Musa Izzanardi Wijanarko, Prasetia Anugrah Pratama, Traci Hong, Ika Idris, Alham Fikri Aji, Derry Wijaya
Abstract: Hate speech poses a significant threat to social harmony. Over the past two years, Indonesia has seen a ten-fold increase in the online hate speech ratio, underscoring the urgent need for effective detection mechanisms. However, progress is hindered by the limited availability of labeled data for Indonesian texts. The condition is even worse for marginalized minorities, such as Shia, LGBTQ, and other ethnic minorities because hate speech is underreported and less understood by detection tools. Furthermore, the lack of accommodation for subjectivity in current datasets compounds this issue. To address this, we introduce IndoToxic2024, a comprehensive Indonesian hate speech and toxicity classification dataset. Comprising 43,692 entries annotated by 19 diverse individuals, the dataset focuses on texts targeting vulnerable groups in Indonesia, specifically during the hottest political event in the country: the presidential election. We establish baselines for seven binary classification tasks, achieving a macro-F1 score of 0.78 with a BERT model (IndoBERTweet) fine-tuned for hate speech classification. Furthermore, we demonstrate how incorporating demographic information can enhance the zero-shot performance of the large language model, gpt-3.5-turbo. However, we also caution that an overemphasis on demographic information can negatively impact the fine-tuned model performance due to data fragmentation.
Authors: Bowen Shen, Kwok-Yan Lam, Feng Li, Li Wang
Abstract: Low earth orbit (LEO) satellite systems play an important role in next generation communication networks due to their ability to provide extensive global coverage with guaranteed communications in remote areas and isolated areas where base stations cannot be cost-efficiently deployed. With the pervasive adoption of LEO satellite systems, especially in the LEO Internet-of-Things (IoT) scenarios, their spectrum resource management requirements have become more complex as a result of massive service requests and high bandwidth demand from terrestrial terminals. For instance, when leasing the spectrum to terrestrial users and controlling the uplink transmit power, satellites collect user data for machine learning purposes, which usually are sensitive information such as location, budget and quality of service (QoS) requirement. To facilitate model training in LEO IoT while preserving the privacy of data, blockchain-driven federated learning (FL) is widely used by leveraging on a fully decentralized architecture. In this paper, we propose a hybrid spectrum pricing and power control framework for LEO IoT by combining blockchain technology and FL. We first design a local deep reinforcement learning algorithm for LEO satellite systems to learn a revenue-maximizing pricing and power control scheme. Then the agents collaborate to form a FL system. We also propose a reputation-based blockchain which is used in the global model aggregation phase of FL. Based on the reputation mechanism, a node is selected for each global training round to perform model aggregation and block generation, which can further enhance the decentralization of the network and guarantee the trust. Simulation tests are conducted to evaluate the performances of the proposed scheme. Our results show the efficiency of finding the maximum revenue scheme for LEO satellite systems while preserving the privacy of each agent.
Authors: Mark Russinovich, Ahmed Salem
Abstract: Growing concerns over the theft and misuse of Large Language Models (LLMs) have heightened the need for effective fingerprinting, which links a model to its original version to detect misuse. In this paper, we define five key properties for a successful fingerprint: Transparency, Efficiency, Persistence, Robustness, and Unforgeability. We introduce a novel fingerprinting framework that provides verifiable proof of ownership while maintaining fingerprint integrity. Our approach makes two main contributions. First, we propose a Chain and Hash technique that cryptographically binds fingerprint prompts with their responses, ensuring no adversary can generate colliding fingerprints and allowing model owners to irrefutably demonstrate their creation. Second, we address a realistic threat model in which instruction-tuned models' output distribution can be significantly altered through meta-prompts. By integrating random padding and varied meta-prompt configurations during training, our method preserves fingerprint robustness even when the model's output style is significantly modified. Experimental results demonstrate that our framework offers strong security for proving ownership and remains resilient against benign transformations like fine-tuning, as well as adversarial attempts to erase fingerprints. Finally, we also demonstrate its applicability to fingerprinting LoRA adapters.
Authors: Terrance Liu, Zhiwei Steven Wu
Abstract: While past works have shown how uncertainty quantification can be applied to large language model (LLM) outputs, the question of whether resulting uncertainty guarantees still hold within sub-groupings of data remains open. In our work, given some long-form text generated by an LLM, we study uncertainty at both the level of individual claims contained within the output (via calibration) and across the entire output itself (via conformal prediction). Using biography generation as a testbed for this study, we derive a set of (demographic) attributes (e.g., whether some text describes a man or woman) for each generation to form such "subgroups" of data. We find that although canonical methods for both types of uncertainty quantification perform well when measuring across the entire dataset, such guarantees break down when examining particular subgroups. Having established this issue, we invoke group-conditional methods for uncertainty quantification -- multicalibration and multivalid conformal prediction -- and find that across a variety of approaches, additional subgroup information consistently improves calibration and conformal prediction within subgroups (while crucially retaining guarantees across the entire dataset). As the problems of calibration, conformal prediction, and their multi-group counterparts have not been extensively explored in the context of long-form text generation, we consider these results to form a benchmark for this setting.
Authors: Jinseong Park, Seungyun Lee, Woojin Jeong, Yujin Choi, Jaewook Lee
Abstract: Time series generation is widely used in real-world applications such as simulation, data augmentation, and hypothesis testing. Recently, diffusion models have emerged as the de facto approach to time series generation, enabling diverse synthesis scenarios. However, the fixed standard-Gaussian diffusion prior may be ill-suited for general time series data, such as temporal order and fixed points. In this paper, we propose TimeBridge, a framework that flexibly synthesizes time series data by using diffusion bridges to learn paths between a chosen prior and the data distribution. We then explore several prior designs tailored to time series synthesis. Our framework covers (i) data- and time-dependent priors for unconditional generation and (ii) scale-preserving priors for conditional generation. Experiments show that our framework with data-driven priors outperforms standard diffusion models on time series generation.
Authors: Samee Arif, Sualeha Farid, Abdul Hameed Azeemi, Awais Athar, Agha Ali Raza
Abstract: This paper presents a novel methodology for generating synthetic Preference Optimization (PO) datasets using multi-model workflows. We evaluate the effectiveness and potential of these workflows in automating and enhancing the dataset generation process. PO dataset generation requires two modules: (1) $\textit{response evaluation}$, and (2) $\textit{response generation}$. In the $\textit{response evaluation}$ module, the responses from Large Language Models (LLMs) are evaluated and ranked - a task typically carried out by human annotators that we automate using LLMs. We assess the response evaluation module in a 2 step process. In step 1, we assess LLMs as evaluators using three distinct prompting strategies. In step 2, we apply the winning prompting strategy to compare the performance of LLM-as-a-Judge, LLMs-as-a-Jury, and LLM Debate. Our evaluation shows that GPT-4o-as-a-Judge is more consistent across all datasets. For the $\textit{response generation}$ module, we use the identified LLM evaluator configuration and compare different configurations of the LLM Feedback Loop. We use the win rate to determine the best multi-model configuration for generation. Experimenting with various configurations, we find that the LLM Feedback Loop, with Llama as the generator and Gemma as the reviewer, achieves a notable 71.8% and 73.8% win rate over single-model Llama and Gemma, respectively. After identifying the best configurations for both modules, we generate our PO datasets using the above pipeline.
Authors: Simon D Angus, Lachlan O'Neill
Abstract: Detecting issue framing in text - how different perspectives approach the same topic - is valuable for social science and policy analysis, yet challenging for automated methods due to subtle linguistic differences. We introduce `paired completion', a novel approach using LLM next-token log probabilities to detect contrasting frames using minimal examples. Through extensive evaluation across synthetic datasets and a human-labeled corpus, we demonstrate that paired completion is a cost-efficient, low-bias alternative to both prompt-based and embedding-based methods, offering a scalable solution for analyzing issue framing in large text collections, especially suited to low-resource settings.
Authors: Hongyang Lei, Xiaolong Cheng, Qi Qin, Dan Wang, Kun Fan, Huazhen Huang, Qingqing Gu, Yetao Wu, Zhonglin Jiang, Yong Chen, Luo Ji
Abstract: Current multimodal learning strategies primarily optimize in the original token space. Such a framework is easy to incorporate with the backbone of pretrained language model, but might result in modality collapse. To alleviate such issues, we leverage the joint embedding predictive architecture (JEPA) on the multimodal tasks, which converts the input embedding into the output embedding space by a predictor and then conducts the cross-modal alignment on the latent space. We implement this predictor by a Multi-Gate Mixture of Experts (MMoE) and name the framework as M3-JEPA, accordingly. The gating function disentangles the modality-specific and shared information and derives information-theoretic optimality. The framework is implemented with both contrastive and regularization loss, and solved by alternative gradient descent (AGD) between different multimodal tasks. By thoroughly designed experiments, we show that M3-JEPA can obtain state-of-the-art performance on different modalities and tasks, generalize to unseen datasets and domains, and is computationally efficient in both training and inference. Our observation suggests that M3-JEPA might become a new basis to self-supervised learning in the open world.
Authors: Sourav Chatterjee, Timothy Sudijono
Abstract: We show that feedforward neural networks with ReLU activation generalize on low complexity data, suitably defined. Given i.i.d.~data generated from a simple programming language, the minimum description length (MDL) feedforward neural network which interpolates the data generalizes with high probability. We define this simple programming language, along with a notion of description length of such networks. We provide several examples on basic computational tasks, such as checking primality of a natural number. For primality testing, our theorem shows the following and more. Suppose that we draw an i.i.d.~sample of $n$ numbers uniformly at random from $1$ to $N$. For each number $x_i$, let $y_i = 1$ if $x_i$ is a prime and $0$ if it is not. Then, the interpolating MDL network accurately answers, with probability $1- O((\ln N)/n)$, whether a newly drawn number between $1$ and $N$ is a prime or not. Note that the network is not designed to detect primes; minimum description learning discovers a network which does so. Extensions to noisy data are also discussed, suggesting that MDL neural network interpolators can demonstrate tempered overfitting.
Authors: Yannis Bendi-Ouis, Dan Dutartre, Xavier Hinaut
Abstract: Since the release of ChatGPT in November 2022, large language models (LLMs) have seen considerable success, including in the open-source community, with many open-weight models available. However, the requirements to deploy such a service are often unknown and difficult to evaluate in advance. To facilitate this process, we conducted numerous tests at the Centre Inria de l'Universit\'e de Bordeaux. In this article, we propose a comparison of the performance of several models of different sizes (mainly Mistral and LLaMa) depending on the available GPUs, using vLLM, a Python library designed to optimize the inference of these models. Our results provide valuable information for private and public groups wishing to deploy LLMs, allowing them to evaluate the performance of different models based on their available hardware. This study thus contributes to facilitating the adoption and use of these large language models in various application domains.
Authors: Maram Assi, Safwat Hassan, Ying Zou
Abstract: The exponential growth of the mobile app market underscores the importance of constant innovation and rapid response to user demands. As user satisfaction is paramount to the success of a mobile application (app), developers typically rely on user reviews, which represent user feedback that includes ratings and comments to identify areas for improvement. However, the sheer volume of user reviews poses challenges in manual analysis, necessitating automated approaches. Existing automated approaches either analyze only the target apps reviews, neglecting the comparison of similar features to competitors or fail to provide suggestions for feature enhancement. To address these gaps, we propose a Large Language Model (LLM)-based Competitive User Review Analysis for Feature Enhancement) (LLM-Cure), an approach powered by LLMs to automatically generate suggestion s for mobile app feature improvements. More specifically, LLM-Cure identifies and categorizes features within reviews by applying LLMs. When provided with a complaint in a user review, LLM-Cure curates highly rated (4 and 5 stars) reviews in competing apps related to the complaint and proposes potential improvements tailored to the target application. We evaluate LLM-Cure on 1,056,739 reviews of 70 popular Android apps. Our evaluation demonstrates that LLM-Cure significantly outperforms the state-of-the-art approaches in assigning features to reviews by up to 13% in F1-score, up to 16% in recall and up to 11% in precision. Additionally, LLM-Cure demonstrates its capability to provide suggestions for resolving user complaints. We verify the suggestions using the release notes that reflect the changes of features in the target mobile app. LLM-Cure achieves a promising average of 73% of the implementation of the provided suggestions.
Authors: Ryan Lagerquist, Galina Chirokova, Robert DeMaria, Mark DeMaria, Imme Ebert-Uphoff
Abstract: Determining the location of a tropical cyclone's (TC) surface circulation center -- "center-fixing" -- is a critical first step in the TC-forecasting process, affecting current/future estimates of track, intensity, and structure. Despite a recent increase in automated center-fixing methods, only one such method (ARCHER-2) is operational, and its best performance is achieved when using microwave or scatterometer data, which are often unavailable. We develop a deep-learning algorithm called GeoCenter; besides a few scalars in the operational Automated Tropical Cyclone Forecasting System, it relies only on geostationary infrared (IR) satellite imagery, which is available for all TC basins at high frequency (10 min) and low latency (< 10 min) during both day and night. GeoCenter ingests an animation (time series) of IR images, including 9 channels at lag times up to 4 hours. The animation is centered at a "first guess" location, offset from the true TC-center location by 48 km on average and sometimes > 100 km; GeoCenter is tasked with correcting this offset. On an independent testing dataset, GeoCenter achieves a mean/median/RMS (root mean square) error of 26.6/22.2/32.4 km for all systems, 24.7/20.8/30.0 km for tropical systems, and 14.6/12.5/17.3 km for category-2--5 hurricanes. These values are similar to ARCHER-2 errors with microwave or scatterometer data, and better than ARCHER-2 errors when only IR data are available. GeoCenter also performs skillful uncertainty quantification, producing a well calibrated ensemble of 150 TC-center locations. Furthermore, all predictors used by GeoCenter are available in real time, which would make GeoCenter easy to implement operationally every 10 min.
Authors: Xiang Hu, Zhihao Teng, Jun Zhao, Wei Wu, Kewei Tu
Abstract: Despite the success of Transformers, handling long contexts remains challenging due to the limited length generalization and quadratic complexity of self-attention. Thus Transformers often require post-training with a larger attention window, significantly increasing computational and memory costs. In this paper, we propose a novel attention mechanism based on dynamic context, Grouped Cross Attention (GCA), which can generalize to 1000 times the pre-training context length while maintaining the ability to access distant information with a constant attention window size. For a given input sequence, we split it into chunks and use each chunk to retrieve top-k relevant past chunks for subsequent text generation. Specifically, unlike most previous works that use an off-the-shelf retriever, our key innovation allows the retriever to learn how to retrieve past chunks that better minimize the auto-regressive loss of subsequent tokens in an end-to-end manner. Such a mechanism accommodates retrieved chunks with a fixed-size attention window to achieve long-range information access, significantly reducing computational and memory costs during training and inference. Experiments show that GCA-based models achieve near-perfect accuracy in passkey retrieval for 16M context lengths, which is 1000 times the training length.
Authors: Zahra Aminiranjbar, Jianan Tang, Qiudan Wang, Shubha Pant, Mahesh Viswanathan
Abstract: The rapid evolution of Large Language Models (LLMs) has transformed them from basic conversational tools into sophisticated entities capable of complex reasoning and decision-making. These advancements have led to the development of specialized LLM-based agents designed for diverse tasks such as coding and web browsing. As these agents become more capable, the need for a robust framework that facilitates global communication and collaboration among them towards advanced objectives has become increasingly critical. Distributed Agents in a Worldwide Network (DAWN) addresses this need by offering a versatile framework that integrates LLM-based agents with traditional software systems, enabling the creation of agentic applications suited for a wide range of use cases. DAWN enables distributed agents worldwide to register and be easily discovered through Gateway Agents. Collaborations among these agents are coordinated by a Principal Agent equipped with reasoning strategies. DAWN offers three operational modes: No-LLM Mode for deterministic tasks, Copilot for augmented decision-making, and LLM Agent for autonomous operations. Additionally, DAWN ensures the safety and security of agent collaborations globally through a dedicated safety, security, and compliance layer, protecting the network against attackers and adhering to stringent security and compliance standards. These features make DAWN a robust network for deploying agent-based applications across various industries.
Authors: Saikat Barua, Mostafizur Rahman, Md Jafor Sadek, Rafiul Islam, Shehenaz Khaled, Md. Shohrab Hossain
Abstract: The principles of automation and innovation serve as foundational elements for advancement in contemporary science and technology. Here, we introduce Pygen, an automation platform designed to empower researchers, technologists, and hobbyists to bring abstract ideas to life as core, usable software tools written in Python. Pygen leverages the immense power of autoregressive large language models to augment human creativity during the ideation, iteration, and innovation process. By combining state-of-the-art language models with open-source code generation technologies, Pygen has significantly reduced the manual overhead of tool development. From a user prompt, Pygen automatically generates Python packages for a complete workflow from concept to package generation and documentation. The findings of our work show that Pygen considerably enhances the researcher's productivity by enabling the creation of resilient, modular, and well-documented packages for various specialized purposes. We employ a prompt enhancement approach to distill the user's package description into increasingly specific and actionable. While being inherently an open-ended task, we have evaluated the generated packages and the documentation using Human Evaluation, LLM-based evaluation, and CodeBLEU, with detailed results in the results section. Furthermore, we documented our results, analyzed the limitations, and suggested strategies to alleviate them. Pygen is our vision of ethical automation, a framework that promotes inclusivity, accessibility, and collaborative development. This project marks the beginning of a large-scale effort towards creating tools where intelligent agents collaborate with humans to improve scientific and technological development substantially. Our code and generated examples are open-sourced at [https://github.com/GitsSaikat/Pygen]
Authors: Erica Coppolillo, Federico Cinus, Marco Minici, Francesco Bonchi, Giuseppe Manco
Abstract: Large Language Models (LLMs) demonstrate significant persuasive capabilities in one-on-one interactions, but their influence within social networks, where interconnected users and complex opinion dynamics pose unique challenges, remains underexplored. This paper addresses the research question: \emph{Can LLMs generate meaningful content that maximizes user engagement on social networks?} To answer this, we propose a pipeline using reinforcement learning with simulated feedback, where the network's response to LLM-generated content (i.e., the reward) is simulated through a formal engagement model. This approach bypasses the temporal cost and complexity of live experiments, enabling an efficient feedback loop between the LLM and the network under study. It also allows to control over endogenous factors such as the LLM's position within the social network and the distribution of opinions on a given topic. Our approach is adaptive to the opinion distribution of the underlying network and agnostic to the specifics of the engagement model, which is embedded as a plug-and-play component. Such flexibility makes it suitable for more complex engagement tasks and interventions in computational social science. Using our framework, we analyze the performance of LLMs in generating social engagement under different conditions, showcasing their full potential in this task. The experimental code is publicly available at https://github.com/mminici/Engagement-Driven-Content-Generation.
URLs: https://github.com/mminici/Engagement-Driven-Content-Generation.
Authors: Xuandong Zhao, Sam Gunn, Miranda Christ, Jaiden Fairoze, Andres Fabrega, Nicholas Carlini, Sanjam Garg, Sanghyun Hong, Milad Nasr, Florian Tramer, Somesh Jha, Lei Li, Yu-Xiang Wang, Dawn Song
Abstract: As the outputs of generative AI (GenAI) techniques improve in quality, it becomes increasingly challenging to distinguish them from human-created content. Watermarking schemes are a promising approach to address the problem of distinguishing between AI and human-generated content. These schemes embed hidden signals within AI-generated content to enable reliable detection. While watermarking is not a silver bullet for addressing all risks associated with GenAI, it can play a crucial role in enhancing AI safety and trustworthiness by combating misinformation and deception. This paper presents a comprehensive overview of watermarking techniques for GenAI, beginning with the need for watermarking from historical and regulatory perspectives. We formalize the definitions and desired properties of watermarking schemes and examine the key objectives and threat models for existing approaches. Practical evaluation strategies are also explored, providing insights into the development of robust watermarking techniques capable of resisting various attacks. Additionally, we review recent representative works, highlight open challenges, and discuss potential directions for this emerging field. By offering a thorough understanding of watermarking in GenAI, this work aims to guide researchers in advancing watermarking methods and applications, and support policymakers in addressing the broader implications of GenAI.
Authors: Jiaming Yu, Le Liang, Chongtao Guo, Ziyang Guo, Shi Jin, Geoffrey Ye Li
Abstract: This paper investigates the use of multi-agent reinforcement learning (MARL) to address distributed channel access in wireless local area networks. In particular, we consider the challenging yet more practical case where the agents heterogeneously adopt value-based or policy-based reinforcement learning algorithms to train the model. We propose a heterogeneous MARL training framework, named QPMIX, which adopts a centralized training with distributed execution paradigm to enable heterogeneous agents to collaborate. Moreover, we theoretically prove the convergence of the proposed heterogeneous MARL method when using the linear value function approximation. Our method maximizes the network throughput and ensures fairness among stations, therefore, enhancing the overall network performance. Simulation results demonstrate that the proposed QPMIX algorithm improves throughput, mean delay, delay jitter, and collision rates compared with conventional carrier-sense multiple access with collision avoidance (CSMA/CA) mechanism in the saturated traffic scenario. Furthermore, the QPMIX algorithm is robust in unsaturated and delay-sensitive traffic scenarios. It coexists well with the conventional CSMA/CA mechanism and promotes cooperation among heterogeneous agents.
Authors: Seyed Mahdi B. Azad, Zahra Padar, Gabriel Kalweit, Joschka Boedecker
Abstract: In this paper, we propose a novel method for learning reward functions directly from offline demonstrations. Unlike traditional inverse reinforcement learning (IRL), our approach decouples the reward function from the learner's policy, eliminating the adversarial interaction typically required between the two. This results in a more stable and efficient training process. Our reward function, called \textit{SR-Reward}, leverages successor representation (SR) to encode a state based on expected future states' visitation under the demonstration policy and transition dynamics. By utilizing the Bellman equation, SR-Reward can be learned concurrently with most reinforcement learning (RL) algorithms without altering the existing training pipeline. We also introduce a negative sampling strategy to mitigate overestimation errors by reducing rewards for out-of-distribution data, thereby enhancing robustness. This strategy inherently introduces a conservative bias into RL algorithms that employ the learned reward. We evaluate our method on the D4RL benchmark, achieving competitive results compared to offline RL algorithms with access to true rewards and imitation learning (IL) techniques like behavioral cloning. Moreover, our ablation studies on data size and quality reveal the advantages and limitations of SR-Reward as a proxy for true rewards.
Authors: Alexander Nikulin, Ilya Zisman, Denis Tarasov, Nikita Lyubaykin, Andrei Polubarov, Igor Kiselev, Vladislav Kurenkov
Abstract: Recently, latent action learning, pioneered by Latent Action Policies (LAPO), have shown remarkable pre-training efficiency on observation-only data, offering potential for leveraging vast amounts of video available on the web for embodied AI. However, prior work has focused on distractor-free data, where changes between observations are primarily explained by ground-truth actions. Unfortunately, real-world videos contain action-correlated distractors that may hinder latent action learning. Using Distracting Control Suite (DCS) we empirically investigate the effect of distractors on latent action learning and demonstrate that LAPO struggle in such scenario. We propose LAOM, a simple LAPO modification that improves the quality of latent actions by 8x, as measured by linear probing. Importantly, we show that providing supervision with ground-truth actions, as few as 2.5% of the full dataset, during latent action learning improves downstream performance by 4.2x on average. Our findings suggest that integrating supervision during Latent Action Models (LAM) training is critical in the presence of distractors, challenging the conventional pipeline of first learning LAM and only then decoding from latent to ground-truth actions.
Authors: Sunny Sanyal, Hayden Prairie, Rudrajit Das, Ali Kavis, Sujay Sanghavi
Abstract: Fine-tuning a pre-trained model on a downstream task often degrades its original capabilities, a phenomenon known as "catastrophic forgetting". This is especially an issue when one does not have access to the data and recipe used to develop the pre-trained model. Under this constraint, most existing methods for mitigating forgetting are inapplicable. To address this challenge, we propose a sample weighting scheme for the fine-tuning data solely based on the pre-trained model's losses. Specifically, we upweight the easy samples on which the pre-trained model's loss is low and vice versa to limit the drift from the pre-trained model. Our approach is orthogonal and yet complementary to existing methods; while such methods mostly operate on parameter or gradient space, we concentrate on the sample space. We theoretically analyze the impact of fine-tuning with our method in a linear setting, showing that it stalls learning in a certain subspace which inhibits overfitting to the target task. We empirically demonstrate the efficacy of our method on both language and vision tasks. As an example, when fine-tuning Gemma 2 2B on MetaMathQA, our method results in only a $0.8\%$ drop in accuracy on GSM8K (another math dataset) compared to standard fine-tuning, while preserving $5.4\%$ more accuracy on the pre-training datasets. Our code is publicly available at https://github.com/sanyalsunny111/FLOW_finetuning .
Authors: Shashwat Goel, Joschka Struber, Ilze Amanda Auzina, Karuna K Chandra, Ponnurangam Kumaraguru, Douwe Kiela, Ameya Prabhu, Matthias Bethge, Jonas Geiping
Abstract: As Language Model (LM) capabilities advance, evaluating and supervising them at scale is getting harder for humans. There is hope that other language models can automate both these tasks, which we refer to as ''AI Oversight''. We study how model similarity affects both aspects of AI oversight by proposing Chance Adjusted Probabilistic Agreement (CAPA): a metric for LM similarity based on overlap in model mistakes. Using CAPA, we first show that LLM-as-a-judge scores favor models similar to the judge, generalizing recent self-preference results. Then, we study training on LM annotations, and find complementary knowledge between the weak supervisor and strong student model plays a crucial role in gains from ''weak-to-strong generalization''. As model capabilities increase, it becomes harder to find their mistakes, and we might defer more to AI oversight. However, we observe a concerning trend -- model mistakes are becoming more similar with increasing capabilities, pointing to risks from correlated failures. Our work underscores the importance of reporting and correcting for model similarity, especially in the emerging paradigm of AI oversight.
Authors: Juyun Wee, Minjae Park, Jaeho Lee
Abstract: Depth pruning aims to reduce the inference cost of a large language model without any hardware-specific complications, by simply removing several less important transformer blocks. However, our empirical findings suggest that the importance of a transformer block may be highly task-dependent -- a block that is crucial for a task can be removed without degrading the accuracy on another task. Based on this observation, we develop a dynamic depth pruning algorithm, coined PuDDing (Prompt-routed Dynamic Depth Pruning), which determines which blocks to omit from the model based on the input prompt. PuDDing operates by training a lightweight router to predict the best omission set among a set of options, where this option set has also been constructed in a data-driven manner. Empirical results on commonsense reasoning benchmarks demonstrate that PuDDing effectively accelerates the inference language models, and achieves better on-task performance than static depth pruning baselines.
Authors: Yeseul Cho, Baekrok Shin, Changmin Kang, Chulhee Yun
Abstract: Recent advances in deep learning rely heavily on massive datasets, leading to substantial storage and training costs. Dataset pruning aims to alleviate this demand by discarding redundant examples. However, many existing methods require training a model with a full dataset over a large number of epochs before being able to prune the dataset, which ironically makes the pruning process more expensive than just training the model on the entire dataset. To overcome this limitation, we introduce a Difficulty and Uncertainty-Aware Lightweight (DUAL) score, which aims to identify important samples from the early training stage by considering both example difficulty and prediction uncertainty. To address a catastrophic accuracy drop at an extreme pruning, we further propose a ratio-adaptive sampling using Beta distribution. Experiments on various datasets and learning scenarios such as image classification with label noise and image corruption, and model architecture generalization demonstrate the superiority of our method over previous state-of-the-art (SOTA) approaches. Specifically, on ImageNet-1k, our method reduces the time cost for pruning to 66% compared to previous methods while achieving a SOTA, specifically 60% test accuracy at a 90% pruning ratio. On CIFAR datasets, the time cost is reduced to just 15% while maintaining SOTA performance.
Authors: Mark Sch\"one, Babak Rahmani, Heiner Kremer, Fabian Falck, Hitesh Ballani, Jannes Gladrow
Abstract: State-space models (SSMs) and transformers dominate the language modeling landscape. However, they are constrained to a lower computational complexity than classical recurrent neural networks (RNNs), limiting their expressivity. In contrast, RNNs lack parallelization during training, raising fundamental questions about the trade off between parallelization and expressivity. We propose implicit SSMs, which iterate a transformation until convergence to a fixed point. Theoretically, we show that implicit SSMs implement the non-linear state-transitions of RNNs. Empirically, we find that only approximate fixed-point convergence suffices, enabling the design of a scalable training curriculum that largely retains parallelization, with full convergence required only for a small subset of tokens. Our approach demonstrates superior state-tracking capabilities on regular languages, surpassing transformers and SSMs. We further scale implicit SSMs to natural language reasoning tasks and pretraining of large-scale language models up to 1.3B parameters on 207B tokens representing, to our knowledge, the largest implicit model trained to date. Notably, our implicit models outperform their explicit counterparts on standard benchmarks. Our code is publicly available at http://github.com/microsoft/implicit_languagemodels .
Authors: Fanxu Meng, Pingzhi Tang, Xiaojuan Tang, Zengwei Yao, Xing Sun, Muhan Zhang
Abstract: In this paper, we present TransMLA, a framework that seamlessly converts any GQA-based pre-trained model into an MLA-based model. Our approach enables direct compatibility with DeepSeek's codebase, allowing these models to fully leverage DeepSeek-specific optimizations such as vLLM and SGlang. By compressing 93% of the KV cache in LLaMA-2-7B, TransMLA achieves a 10.6x inference speedup at an 8K context length while preserving meaningful output quality. Additionally, the model requires only 6 billion tokens for fine-tuning to regain performance on par with the original across multiple benchmarks. TransMLA offers a practical solution for migrating GQA-based models to the MLA structure. When combined with DeepSeek's advanced features, such as FP8 quantization and Multi-Token Prediction, even greater inference acceleration can be realized.
Authors: Jinda Xu, Yuhao Song, Daming Wang, Weiwei Zhao, Minghua Chen, Kangliang Chen, Qinya Li
Abstract: In an era overwhelmed by vast amounts of data, the effective curation of web-crawl datasets is essential for optimizing model performance. This paper tackles the challenges associated with the unstructured and heterogeneous nature of such datasets. Traditional heuristic curation methods often inadequately capture complex features, resulting in biases and the exclusion of relevant data. We introduce an advanced, learning-driven approach, Ensemble Curation Of DAta ThroUgh Multimodal Operators (EcoDatum), incorporating a novel quality-guided deduplication method to ensure balanced feature distributions. EcoDatum strategically integrates various unimodal and multimodal data curation operators within a weak supervision ensemble framework, utilizing automated optimization to score each data point effectively. EcoDatum, which significantly improves the data curation quality and efficiency, outperforms existing state-of-the-art (SOTA) techniques, ranked 1st on the DataComp leaderboard, with an average performance score of 0.182 across 38 diverse evaluation datasets. This represents a 28% improvement over the DataComp baseline method, demonstrating its effectiveness in improving dataset curation and model training efficiency.
Authors: Albina Klepach, Alexander Nikulin, Ilya Zisman, Denis Tarasov, Alexander Derevyagin, Andrei Polubarov, Nikita Lyubaykin, Vladislav Kurenkov
Abstract: Leveraging vast amounts of unlabeled internet video data for embodied AI is currently bottlenecked by the lack of action labels and the presence of action-correlated visual distractors. Although recent latent action policy optimization (LAPO) has shown promise in inferring proxy-action labels from visual observations, its performance degrades significantly when distractors are present. To address this limitation, we propose a novel object-centric latent action learning framework that centers on objects rather than pixels. We leverage self-supervised object-centric pretraining to disentangle action-related and distracting dynamics. This allows LAPO to focus on task-relevant interactions, resulting in more robust proxy-action labels, enabling better imitation learning and efficient adaptation of the agent with just a few action-labeled trajectories. We evaluated our method in eight visually complex tasks across the Distracting Control Suite (DCS) and Distracting MetaWorld (DMW). Our results show that object-centric pretraining mitigates the negative effects of distractors by 50%, as measured by downstream task performance: average return (DCS) and success rate (DMW).
Authors: Phaphontee Yamchote, Saw Nay Htet Win, Chainarong Amornbunchornvej, Thanapon Noraset
Abstract: Feature interaction is crucial in predictive machine learning models, as it captures the relationships between features that influence model performance. In this work, we focus on pairwise interactions and investigate their importance in constructing feature graphs for Graph Neural Networks (GNNs). We leverage existing GNN models and tools to explore the relationship between feature graph structures and their effectiveness in modeling interactions. Through experiments on synthesized datasets, we uncover that edges between interacting features are important for enabling GNNs to model feature interactions effectively. We also observe that including non-interaction edges can act as noise, degrading model performance. Furthermore, we provide theoretical support for sparse feature graph selection using the Minimum Description Length (MDL) principle. We prove that feature graphs retaining only necessary interaction edges yield a more efficient and interpretable representation than complete graphs, aligning with Occam's Razor. Our findings offer both theoretical insights and practical guidelines for designing feature graphs that improve the performance and interpretability of GNN models.
Authors: Mark Russinovich, Ahmed Salem
Abstract: Recent copyright agreements between AI companies and content creators underscore the need for fine-grained control over language models' ability to reproduce copyrighted text. Existing defenses-ranging from aggressive unlearning to simplistic output filters-either sacrifice model utility or inadequately address verbatim leakage. We introduce Obliviate, a lightweight post-training method that surgically suppresses exact reproduction of specified sequences while preserving semantic understanding. Obliviate first identifies memorized passages and then, for each target token, minimally adjusts the model's output distribution via a Kullback-Leibler divergence penalty to drive down the probability of exact reproduction. Simultaneously, we enforce a consistency loss on non-target tokens to retain the model's fluency and task performance. We evaluate Obliviate on four popular 6-8B-parameter models (LLaMA-3.1, LLaMA-3.1-Instruct, Qwen-2.5, and Yi-1.5) using synthetic memorization benchmarks and organic copyrighted excerpts (e.g., Moby Dick, Frankenstein, Alice in Wonderland and Les Miserables). Across all settings, Obliviate reduces verbatim recall by two orders of magnitude (e.g., from hundreds of words to fewer than 12) while degrading downstream accuracy by at most 1% on HellaSwag, MMLU, TruthfulQA, and Winogrande. Furthermore, we benchmark Obliviate aganist different unlearning and copyright techniques using the MUSE and CoTaEval benchmarks. These results position Obliviate as a practical, high-fidelity solution for copyright compliance in deployed LLMs.
Authors: MohammadHossein Rezaei, Yicheng Fu, Phil Cuvin, Caleb Ziems, Yanzhe Zhang, Hao Zhu, Diyi Yang
Abstract: Human activity is moderated by norms; however, supervision for normative reasoning is sparse, particularly where norms are physically- or socially-grounded. We thus present EGONORMIA $\|\epsilon\|$, comprising 1,853 (200 for EGONORMIA-verified) multiple choice questions (MCQs) grounded within egocentric videos of human interactions, enabling the evaluation and improvement of normative reasoning in vision-language models (VLMs). EGONORMIA spans seven norm categories: safety, privacy, proxemics, politeness, cooperation, coordination/proactivity, and communication/legibility. To compile this dataset at scale, we propose a novel pipeline to generate grounded MCQs from raw egocentric video. Our work demonstrates that current state-of-the-art VLMs lack robust grounded norm understanding, scoring a maximum of 54% on EGONORMIA and 65% on EGONORMIA-verified, with performance across norm categories indicating significant risks of safety and privacy when VLMs are used in real-world agents. We additionally explore methods for improving normative understanding, demonstrating that a naive retrieval-based generation (RAG) method using EGONORMIA can enhance normative reasoning in VLMs.
Authors: Metod Jazbec, Eliot Wong-Toi, Guoxuan Xia, Dan Zhang, Eric Nalisnick, Stephan Mandt
Abstract: Diffusion models have recently driven significant breakthroughs in generative modeling. While state-of-the-art models produce high-quality samples on average, individual samples can still be low quality. Detecting such samples without human inspection remains a challenging task. To address this, we propose a Bayesian framework for estimating generative uncertainty of synthetic samples. We outline how to make Bayesian inference practical for large, modern generative models and introduce a new semantic likelihood (evaluated in the latent space of a feature extractor) to address the challenges posed by high-dimensional sample spaces. Through our experiments, we demonstrate that the proposed generative uncertainty effectively identifies poor-quality samples and significantly outperforms existing uncertainty-based methods. Notably, our Bayesian framework can be applied post-hoc to any pretrained diffusion or flow matching model (via the Laplace approximation), and we propose simple yet effective techniques to minimize its computational overhead during sampling.
Authors: Antonio M. Mercado-Mart\'inez, Beatriz Soret, Antonio Jurado-Navas
Abstract: The Agile Earth Observation Satellite Scheduling Problem (AEOSSP) entails finding the subset of observation targets to be scheduled along the satellite's orbit while meeting operational constraints of time, energy and memory. The problem of deciding what and when to observe is inherently complex, and becomes even more challenging when considering several issues that compromise the quality of the captured images, such as cloud occlusion, atmospheric turbulence, and image resolution. This paper presents a Deep Reinforcement Learning (DRL) approach for addressing the AEOSSP with time-dependent profits, integrating these three factors to optimize the use of energy and memory resources. The proposed method involves a dual decision-making process: selecting the sequence of targets and determining the optimal observation time for each. Our results demonstrate that the proposed algorithm reduces the capture of images that fail to meet quality requirements by > 60% and consequently decreases energy waste from attitude maneuvers by up to 78%, all while maintaining strong observation performance.
Authors: Jiho Jin, Woosung Kang, Junho Myung, Alice Oh
Abstract: Measuring social bias in large language models (LLMs) is crucial, but existing bias evaluation methods struggle to assess bias in long-form generation. We propose a Bias Benchmark for Generation (BBG), an adaptation of the Bias Benchmark for QA (BBQ), designed to evaluate social bias in long-form generation by having LLMs generate continuations of story prompts. Building our benchmark in English and Korean, we measure the probability of neutral and biased generations across ten LLMs. We also compare our long-form story generation evaluation results with multiple-choice BBQ evaluation, showing that the two approaches produce inconsistent results.
Authors: Xiaoxiao Liu, Qingying Xiao, Junying Chen, Xiangyi Feng, Xiangbo Wu, Bairui Zhang, Xiang Wan, Jian Chang, Guangjun Yu, Yan Hu, Benyou Wang
Abstract: Large language models (LLMs) are increasingly applied to outpatient referral tasks across healthcare systems. However, there is a lack of standardized evaluation criteria to assess their effectiveness, particularly in dynamic, interactive scenarios. In this study, we systematically examine the capabilities and limitations of LLMs in managing tasks within Intelligent Outpatient Referral (IOR) systems and propose a comprehensive evaluation framework specifically designed for such systems. This framework comprises two core tasks: static evaluation, which focuses on evaluating the ability of predefined outpatient referrals, and dynamic evaluation, which evaluates capabilities of refining outpatient referral recommendations through iterative dialogues. Our findings suggest that LLMs offer limited advantages over BERT-like models, but show promise in asking effective questions during interactive dialogues.
Authors: Sumin In, Youngdong Jang, Utae Jeong, MinHyuk Jang, Hyeongcheol Park, Eunbyung Park, Sangpil Kim
Abstract: 3D Gaussian Splatting (3DGS) is increasingly adopted in various academic and commercial applications due to its real-time and high-quality rendering capabilities, emphasizing the growing need for copyright protection technologies for 3DGS. However, the large model size of 3DGS requires developing efficient compression techniques. This highlights the necessity of an integrated framework that addresses copyright protection and data compression for 3D content. Nevertheless, existing 3DGS watermarking methods significantly degrade watermark performance under 3DGS compression methods, particularly quantization-based approaches that achieve superior compression performance. To ensure reliable watermark detection under compression, we propose a compression-tolerant anchor-based 3DGS watermarking, which preserves watermark integrity and rendering quality. This is achieved by introducing anchor-based 3DGS watermarking. We embed the watermark into the anchor attributes, particularly the anchor feature, to enhance security and rendering quality. We also propose a quantization distortion layer that injects quantization noise during training, preserving the watermark after quantization-based compression. Moreover, we employ a frequency-aware anchor growing strategy that improves rendering quality and watermark performance by effectively identifying Gaussians in high-frequency regions. Extensive experiments demonstrate that our proposed method preserves the watermark even under compression and maintains high rendering quality.
Authors: Chi Han, Heng Ji
Abstract: Most written natural languages are composed of sequences of words and sentences. Similar to humans, large language models (LLMs) exhibit flexibility in handling textual positions - a phenomenon we term position generalization. They can understand texts with position perturbations and generalize to longer texts than those encountered during training with the latest techniques. These phenomena suggest that LLMs handle positions tolerantly, but how LLMs computationally process positional relevance remains largely unexplored. This work connects the linguistic phenomenon with LLMs' computational mechanisms. We show how LLMs enforce certain computational mechanisms for the aforementioned tolerance in position perturbations. Despite the complex design of the self-attention mechanism, this work reveals that LLMs learn a counterintuitive disentanglement of attention logits. Their values show a 0.959 linear correlation with an approximation of the arithmetic sum of positional relevance and semantic importance. Furthermore, we identify a prevalent pattern in intermediate features, which we prove theoretically enables this effect. The pattern, which is different from how randomly initialized parameters would behave, suggests that it is a learned behavior rather than a natural result of the model architecture. Based on these findings, we provide computational explanations and criteria for LLMs' position flexibilities. This work takes a pioneering step in linking position generalization with modern LLMs' internal mechanisms.
Authors: Wei Fang, Yang Zhang, Kaizhi Qian, James Glass, Yada Zhu
Abstract: Large language models (LLMs) are increasingly integrated with specialized external tools, yet many tasks demand zero-shot tool usage with minimal or noisy documentation. Existing solutions rely on manual rewriting or labeled data for validation, making them inapplicable in true zero-shot settings. To address these challenges, we propose PLAY2PROMPT, an automated framework that systematically "plays" with each tool to explore its input-output behaviors. Through this iterative trial-and-error process, PLAY2PROMPT refines tool documentation and generates usage examples without any labeled data. These examples not only guide LLM inference but also serve as validation to further enhance tool utilization. Extensive experiments on real-world tasks demonstrate that PLAY2PROMPT significantly improves zero-shot tool performance across both open and closed models, offering a scalable and effective solution for domain-specific tool integration.
Authors: Kar Balan, Andrew Gilbert, John Collomosse
Abstract: The rise of Generative AI (GenAI) has sparked significant debate over balancing the interests of creative rightsholders and AI developers. As GenAI models are trained on vast datasets that often include copyrighted material, questions around fair compensation and proper attribution have become increasingly urgent. To address these challenges, this paper proposes a framework called Content ARCs (Authenticity, Rights, Compensation). By combining open standards for provenance and dynamic licensing with data attribution, and decentralized technologies, Content ARCs create a mechanism for managing rights and compensating creators for using their work in AI training. We characterize several nascent works in the AI data licensing space within Content ARCs and identify where challenges remain to fully implement the end-to-end framework.
Authors: Yike Yuan, Ziyu Wang, Zihao Huang, Defa Zhu, Xun Zhou, Jingyi Yu, Qiyang Min
Abstract: Diffusion models have emerged as mainstream framework in visual generation. Building upon this success, the integration of Mixture of Experts (MoE) methods has shown promise in enhancing model scalability and performance. In this paper, we introduce Race-DiT, a novel MoE model for diffusion transformers with a flexible routing strategy, Expert Race. By allowing tokens and experts to compete together and select the top candidates, the model learns to dynamically assign experts to critical tokens. Additionally, we propose per-layer regularization to address challenges in shallow layer learning, and router similarity loss to prevent mode collapse, ensuring better expert utilization. Extensive experiments on ImageNet validate the effectiveness of our approach, showcasing significant performance gains while promising scaling properties.
Authors: Reza Sameni
Abstract: We study the geometry of Receiver Operating Characteristic (ROC) and Precision-Recall (PR) curves in binary classification problems. The key finding is that many of the most commonly used binary classification metrics are merely functions of the composition function $G := F_p \circ F_n^{-1}$, where $F_p(\cdot)$ and $F_n(\cdot)$ are the class-conditional cumulative distribution functions of the classifier scores in the positive and negative classes, respectively. This geometric perspective facilitates the selection of operating points, understanding the effect of decision thresholds, and comparison between classifiers. It also helps explain how the shapes and geometry of ROC/PR curves reflect classifier behavior, providing objective tools for building classifiers optimized for specific applications with context-specific constraints. We further explore the conditions for classifier dominance, present analytical and numerical examples demonstrating the effects of class separability and variance on ROC and PR geometries, and derive a link between the positive-to-negative class leakage function $G(\cdot)$ and the Kullback--Leibler divergence. The framework highlights practical considerations, such as model calibration, cost-sensitive optimization, and operating point selection under real-world capacity constraints, enabling more informed approaches to classifier deployment and decision-making.
Authors: Wei Xiong, Jiarui Yao, Yuhui Xu, Bo Pang, Lei Wang, Doyen Sahoo, Junnan Li, Nan Jiang, Tong Zhang, Caiming Xiong, Hanze Dong
Abstract: Reinforcement learning (RL) has become a prevailing approach for fine-tuning large language models (LLMs) on complex reasoning tasks. Among recent methods, GRPO stands out for its empirical success in training models such as DeepSeek-R1, yet the sources of its effectiveness remain poorly understood. In this work, we revisit GRPO from a reinforce-like algorithm perspective and analyze its core components. Surprisingly, we find that a simple rejection sampling baseline, RAFT, which trains only on positively rewarded samples, yields competitive performance than GRPO and PPO. Our ablation studies reveal that GRPO's main advantage arises from discarding prompts with entirely incorrect responses, rather than from its reward normalization. Motivated by this insight, we propose Reinforce-Rej, a minimal extension of policy gradient that filters both entirely incorrect and entirely correct samples. Reinforce-Rej improves KL efficiency and stability, serving as a lightweight yet effective alternative to more complex RL algorithms. We advocate RAFT as a robust and interpretable baseline, and suggest that future advances should focus on more principled designs for incorporating negative samples, rather than relying on them indiscriminately. Our findings provide guidance for future work in reward-based LLM post-training.
Authors: Cheng-Yen Hsieh, Xinyou Wang, Daiheng Zhang, Dongyu Xue, Fei Ye, Shujian Huang, Zaixiang Zheng, Quanquan Gu
Abstract: Multimodal protein language models (PLMs) integrate sequence and token-based structural information, serving as a powerful foundation for protein modeling, generation, and design. However, the reliance on tokenizing 3D structures into discrete tokens causes substantial loss of fidelity about fine-grained structural details and correlations. In this paper, we systematically elucidate the design space of multimodal PLMs to overcome their limitations. We identify tokenization loss and inaccurate structure token predictions by the PLMs as major bottlenecks. To address these, our proposed design space covers improved generative modeling, structure-aware architectures and representation learning, and data exploration. Our advancements approach finer-grained supervision, demonstrating that token-based multimodal PLMs can achieve robust structural modeling. The effective design methods dramatically improve the structure generation diversity, and notably, folding abilities of our 650M model by reducing the RMSD from 5.52 to 2.36 on PDB testset, even outperforming 3B baselines and on par with the specialized folding models. Project page and code: https://bytedance.github.io/dplm/dplm-2.1/.
Authors: William Corrias, Fabio De Gaspari, Dorjan Hitaj, Luigi V. Mancini
Abstract: Recent advances in generative models have led to their application in password guessing, with the aim of replicating the complexity, structure, and patterns of human-created passwords. Despite their potential, inconsistencies and inadequate evaluation methodologies in prior research have hindered meaningful comparisons and a comprehensive, unbiased understanding of their capabilities. This paper introduces MAYA, a unified, customizable, plug-and-play benchmarking framework designed to facilitate the systematic characterization and benchmarking of generative password-guessing models in the context of trawling attacks. Using MAYA, we conduct a comprehensive assessment of six state-of-the-art approaches, which we re-implemented and adapted to ensure standardization. Our evaluation spans eight real-world password datasets and covers an exhaustive set of advanced testing scenarios, totaling over 15,000 compute hours. Our findings indicate that these models effectively capture different aspects of human password distribution and exhibit strong generalization capabilities. However, their effectiveness varies significantly with long and complex passwords. Through our evaluation, sequential models consistently outperform other generative architectures and traditional password-guessing tools, demonstrating unique capabilities in generating accurate and complex guesses. Moreover, the diverse password distributions learned by the models enable a multi-model attack that outperforms the best individual model. By releasing MAYA, we aim to foster further research, providing the community with a new tool to consistently and reliably benchmark generative password-guessing models. Our framework is publicly available at https://github.com/williamcorrias/MAYA-Password-Benchmarking.
URLs: https://github.com/williamcorrias/MAYA-Password-Benchmarking.
Authors: Alan Lee, Harry Tong
Abstract: We propose reinforcement learning (RL) strategies tailored for reasoning in large language models (LLMs) under strict memory and compute limits, with a particular focus on compatibility with LoRA fine-tuning. Building on early policy gradient methods with baseline subtraction, we design critic-free methods that operate on a small, informative subset of output tokens to reduce memory usage and stabilize training. We introduce S-GRPO, a stochastic variant of Group Relative Policy Optimization, and T-SPMO, a token-level prefix matching approach for fine-grained credit assignment. Applied to Qwen2-1.5B, our methods raise accuracy on the SVAMP benchmark from 46% to over 70% and show strong performance on multi-digit multiplication. Surprisingly, full-token GRPO under LoRA fails to improve over the base model, suggesting that selective token-level optimization may act as an implicit regularizer in low-parameter training regimes.
Authors: Trisanth Srinivasan, Santosh Patapati
Abstract: Robust navigation in diverse environments and domains requires both accurate state estimation and transparent decision making. We present PhysNav-DG, a novel framework that integrates classical sensor fusion with the semantic power of vision-language models. Our dual-branch architecture predicts navigation actions from multi-sensor inputs while simultaneously generating detailed chain-of-thought explanations. A modified Adaptive Kalman Filter dynamically adjusts its noise parameters based on environmental context. It leverages several streams of raw sensor data along with semantic insights from models such as LLaMA 3.2 11B and BLIP-2. To evaluate our approach, we introduce the MD-NEX Benchmark, a novel multi-domain dataset that unifies indoor navigation, autonomous driving, and social navigation tasks with ground-truth actions and human-validated explanations. Extensive experiments and ablations show that PhysNav-DG improves navigation success rates by over 20% and achieves high efficiency, with explanations that are both highly grounded and clear. This work connects high-level semantic reasoning and geometric planning for safer and more trustworthy autonomous systems.
Authors: Xiaoyu Wang, Yue Zhao, Qingqing Gu, Zhonglin Jiang, Xiaokai Chen, Yong Chen, Luo Ji
Abstract: Emotional support conversation (ESC) aims to alleviate the emotional distress of individuals through effective conversations. Although large language models (LLMs) have obtained remarkable progress on ESC, most of these studies might not define the diagram from the state model perspective, therefore providing a suboptimal solution for long-term satisfaction. To address such an issue, we leverage the Q-learning on LLMs, and propose a framework called straQ*. Our framework allows a plug-and-play LLM to bootstrap the planning during ESC, determine the optimal strategy based on long-term returns, and finally guide the LLM to response. Substantial experiments on ESC datasets suggest that straQ* outperforms many baselines, including direct inference, self-refine, chain of thought, finetuning, and finite state machines.
Authors: Saikat Barua, Mostafizur Rahman, Shehenaz Khaled, Md Jafor Sadek, Rafiul Islam, Shahnewaz Siddique
Abstract: The emergence of hybrid quantum-classical machine learning (HQML) models opens new horizons of computational intelligence but their fundamental complexity frequently leads to black box behavior that undermines transparency and reliability in their application. Although XAI for quantum systems still in its infancy, a major research gap is evident in robust global and local explainability approaches that are designed for HQML architectures that employ quantized feature encoding followed by classical learning. The gap is the focus of this work, which introduces QuXAI, an framework based upon Q-MEDLEY, an explainer for explaining feature importance in these hybrid systems. Our model entails the creation of HQML models incorporating quantum feature maps, the use of Q-MEDLEY, which combines feature based inferences, preserving the quantum transformation stage and visualizing the resulting attributions. Our result shows that Q-MEDLEY delineates influential classical aspects in HQML models, as well as separates their noise, and competes well against established XAI techniques in classical validation settings. Ablation studies more significantly expose the virtues of the composite structure used in Q-MEDLEY. The implications of this work are critically important, as it provides a route to improve the interpretability and reliability of HQML models, thus promoting greater confidence and being able to engage in safer and more responsible use of quantum-enhanced AI technology. Our code and experiments are open-sourced at: https://github.com/GitsSaikat/QuXAI
Authors: Shihao Zhang, Haoyu Zhang, Ian Colbert, Rayan Saab
Abstract: We introduce Qronos -- a new state-of-the-art post-training quantization algorithm that sequentially rounds and updates neural network weights. Qronos not only explicitly corrects errors due to both weight and activation quantization, but also errors resulting from quantizing previous layers. Our iterative algorithm is based on an interpretable and disciplined optimization framework that subsumes and surpasses existing data-driven approaches. At each step, Qronos alternates between error correction and diffusion via optimal update rules. Importantly, we prove that Qronos admits an efficient implementation that uses the Cholesky decomposition for solving least-squares problems. We also demonstrate that Qronos is compatible with existing transformation techniques such as Hadamard-based incoherence processing and weight-activation scaling equalization, among others. We evaluate Qronos using recent autoregressive language generation models in the Llama3 family; Qronos consistently outperforms previous state-of-the-art adaptive rounding methods when quantizing the weights, activations, and/or KV caches.
Authors: Mingyuan Wu, Jingcheng Yang, Jize Jiang, Meitang Li, Kaizhuo Yan, Hanchao Yu, Minjia Zhang, Chengxiang Zhai, Klara Nahrstedt
Abstract: Reinforcement Learning Finetuning (RFT) has significantly advanced the reasoning capabilities of large language models (LLMs) by enabling long chains of thought, self-correction, and effective tool use. While recent works attempt to extend RFT to vision-language models (VLMs), these efforts largely produce text-only reasoning conditioned on static image inputs, falling short of true multimodal reasoning in the response. In contrast, test-time methods like Visual Sketchpad incorporate visual steps but lack training mechanisms. We introduce VTool-R1, the first framework that trains VLMs to generate multimodal chains of thought by interleaving text and intermediate visual reasoning steps. VTool-R1 integrates Python-based visual editing tools into the RFT process, enabling VLMs to learn when and how to generate visual reasoning steps that benefit final reasoning. Trained with outcome-based rewards tied to task accuracy, our approach elicits strategic visual tool use for reasoning without relying on process-based supervision. Experiments on structured visual question answering over charts and tables show that VTool-R1 enhances reasoning performance by teaching VLMs to "think with images" and generate multimodal chain of thoughts with tools.
Authors: Elija Perrier
Abstract: AIXI is a widely studied model of artificial general intelligence (AGI) based upon principles of induction and reinforcement learning. However, AIXI is fundamentally classical in nature - as are the environments in which it is modelled. Given the universe is quantum mechanical in nature and the exponential overhead required to simulate quantum mechanical systems classically, the question arises as to whether there are quantum mechanical analogues of AIXI. To address this question, we extend the framework to quantum information and present Quantum AIXI (QAIXI). We introduce a model of quantum agent/environment interaction based upon quantum and classical registers and channels, showing how quantum AIXI agents may take both classical and quantum actions. We formulate the key components of AIXI in quantum information terms, extending previous research on quantum Kolmogorov complexity and a QAIXI value function. We discuss conditions and limitations upon quantum Solomonoff induction and show how contextuality fundamentally affects QAIXI models.
Authors: Shifeng Xie, Aref Einizade, Jhony H. Giraldo
Abstract: Graph Representation Learning (GRL) is a fundamental task in machine learning, aiming to encode high-dimensional graph-structured data into low-dimensional vectors. Self-Supervised Learning (SSL) methods are widely used in GRL because they can avoid expensive human annotation. In this work, we propose a novel Subgraph Gaussian Embedding Contrast (SubGEC) method. Our approach introduces a subgraph Gaussian embedding module, which adaptively maps subgraphs to a structured Gaussian space, ensuring the preservation of input subgraph characteristics while generating subgraphs with a controlled distribution. We then employ optimal transport distances, more precisely the Wasserstein and Gromov-Wasserstein distances, to effectively measure the similarity between subgraphs, enhancing the robustness of the contrastive learning process. Extensive experiments across multiple benchmarks demonstrate that \method~outperforms or presents competitive performance against state-of-the-art approaches. Our findings provide insights into the design of SSL methods for GRL, emphasizing the importance of the distribution of the generated contrastive pairs.
Authors: Thinh Pham, Nguyen Nguyen, Pratibha Zunjare, Weiyuan Chen, Yu-Min Tseng, Tu Vu
Abstract: We introduce SealQA, a new challenge benchmark for evaluating SEarch-Augmented Language models on fact-seeking questions where web search yields conflicting, noisy, or unhelpful results. SealQA comes in three flavors: (1) Seal-0 (main) and (2) Seal-Hard, which assess factual accuracy and reasoning capabilities, with Seal-0 focusing on the most challenging questions where chat models (e.g., GPT-4.1) typically achieve near-zero accuracy; and (3) LongSeal, which extends SealQA to test long-context, multi-document reasoning in "needle-in-a-haystack" settings. Our evaluation reveals critical limitations in current models: Even frontier LLMs perform poorly across all SealQA flavors. On Seal-0, frontier agentic models equipped with tools like o3 and o4-mini achieve only 17.1% and 6.3% accuracy, respectively, at their best reasoning efforts. We find that advanced reasoning models such as DeepSeek-R1-671B and o3-mini are highly vulnerable to noisy search results. Notably, increasing test-time compute does not yield reliable gains across o3-mini, o4-mini, and o3, with performance often plateauing or even declining early. Additionally, while recent models are less affected by the "lost-in-the-middle" issue, they still fail to reliably identify relevant documents in LongSeal when faced with numerous distractors. To facilitate future work, we release SealQA at huggingface.co/datasets/vtllms/sealqa.
Authors: Yulei Qin, Gang Li, Zongyi Li, Zihan Xu, Yuchen Shi, Zhekai Lin, Xiao Cui, Ke Li, Xing Sun
Abstract: Existing large language models (LLMs) face challenges of following complex instructions, especially when multiple constraints are present and organized in paralleling, chaining, and branching structures. One intuitive solution, namely chain-of-thought (CoT), is expected to universally improve capabilities of LLMs. However, we find that the vanilla CoT exerts a negative impact on performance due to its superficial reasoning pattern of simply paraphrasing the instructions. It fails to peel back the compositions of constraints for identifying their relationship across hierarchies of types and dimensions. To this end, we propose a systematic method to boost LLMs in dealing with complex instructions via incentivizing reasoning for test-time compute scaling. First, we stem from the decomposition of complex instructions under existing taxonomies and propose a reproducible data acquisition method. Second, we exploit reinforcement learning (RL) with verifiable rule-centric reward signals to cultivate reasoning specifically for instruction following. We address the shallow, non-essential nature of reasoning under complex instructions via sample-wise contrast for superior CoT enforcement. We also exploit behavior cloning of experts to facilitate steady distribution shift from fast-thinking LLMs to skillful reasoners. Extensive evaluations on seven comprehensive benchmarks confirm the validity of the proposed method, where a 1.5B LLM achieves 11.74% gains with performance comparable to a 8B LLM. Codes and data are available at https://github.com/yuleiqin/RAIF.
Authors: Shuai Wang, Yinan Yu
Abstract: While Large Language Models (LLMs) excel at many natural language processing tasks, they often suffer from factual inaccuracies in knowledge-intensive scenarios. Integrating external knowledge resources, particularly knowledge graphs (KGs), provides a transparent and updatable foundation for more reliable reasoning. Knowledge Base Question Answering (KBQA), which queries and reasons over KGs, is central to this effort, especially for complex, multi-hop queries. However, multi-hop reasoning poses two key challenges: (1)~maintaining coherent reasoning paths, and (2)~avoiding prematurely discarding critical multi-hop connections. To address these issues, we introduce iQUEST, a question-guided KBQA framework that iteratively decomposes complex queries into simpler sub-questions, ensuring a structured and focused reasoning trajectory. Additionally, we integrate a Graph Neural Network (GNN) to look ahead and incorporate 2-hop neighbor information at each reasoning step. This dual approach strengthens the reasoning process, enabling the model to explore viable paths more effectively. Detailed experiments demonstrate the consistent improvement delivered by iQUEST across four benchmark datasets and four LLMs.
Authors: Zeng Wang, Minghao Shao, Rupesh Karn, Likhitha Mankali, Jitendra Bhandari, Ramesh Karri, Ozgur Sinanoglu, Muhammad Shafique, Johann Knechtel
Abstract: Large Language Models (LLMs) offer transformative capabilities for hardware design automation, particularly in Verilog code generation. However, they also pose significant data security challenges, including Verilog evaluation data contamination, intellectual property (IP) design leakage, and the risk of malicious Verilog generation. We introduce SALAD, a comprehensive assessment that leverages machine unlearning to mitigate these threats. Our approach enables the selective removal of contaminated benchmarks, sensitive IP and design artifacts, or malicious code patterns from pre-trained LLMs, all without requiring full retraining. Through detailed case studies, we demonstrate how machine unlearning techniques effectively reduce data security risks in LLM-aided hardware design.
Authors: Guohang Zhuang, Weixi Song, Jinyang Huang, Chenwei Yang, Wanli OuYang, Yan Lu
Abstract: With the rapid development of space exploration, space debris has attracted more attention due to its potential extreme threat, leading to the need for real-time and accurate debris tracking. However, existing methods are mainly based on traditional signal processing, which cannot effectively process the complex background and dense space debris. In this paper, we propose a deep learning-based Space Debris Tracking Network~(SDT-Net) to achieve highly accurate debris tracking. SDT-Net effectively represents the feature of debris, enhancing the efficiency and stability of end-to-end model learning. To train and evaluate this model effectively, we also produce a large-scale dataset Space Debris Tracking Dataset (SDTD) by a novel observation-based data simulation scheme. SDTD contains 18,040 video sequences with a total of 62,562 frames and covers 250,000 synthetic space debris. Extensive experiments validate the effectiveness of our model and the challenging of our dataset. Furthermore, we test our model on real data from the Antarctic Station, achieving a MOTA score of 70.6%, which demonstrates its strong transferability to real-world scenarios. Our dataset and code will be released soon.
Authors: Alex Pan, Mary-Anne Williams
Abstract: The dominant way of judging Large Language Models (LLMs) has been to ask how well they can recall explicit facts from very long inputs. While today's best models achieve near perfect recall, this masks a harder skill: performing multi-step reasoning and tracking intermediate state that never appears verbatim. We introduce Verbose ListOps (VLO), a benchmark that embeds deterministic ListOps computations inside narrative camouflage and, crucially, allows step-level evaluation of every intermediate result. Experiments show that models which solve raw ListOps with approximately 100% accuracy collapse on VLO after only 10,000 tokens. By exposing where a model's reasoning chain first diverges, VLO moves assessment beyond sheer context length and toward genuine comprehension. VLO's generation pipeline is task-agnostic: it can weave any deterministically verifiable reasoning schema -- arithmetic, symbolic, abductive, inductive or defeasible -- into narrative form. This makes VLO a reusable test-bed for the next wave of reasoning-centric model designs, not merely those with step-explicit scaffolds.
Authors: Baihe Huang, Shanda Li, Tianhao Wu, Yiming Yang, Ameet Talwalkar, Kannan Ramchandran, Michael I. Jordan, Jiantao Jiao
Abstract: Test-time scaling paradigms have significantly advanced the capabilities of large language models (LLMs) on complex tasks. Despite their empirical success, theoretical understanding of the sample efficiency of various test-time strategies -- such as self-consistency, best-of-$n$, and self-correction -- remains limited. In this work, we first establish a separation result between two repeated sampling strategies: self-consistency requires $\Theta(1/\Delta^2)$ samples to produce the correct answer, while best-of-$n$ only needs $\Theta(1/\Delta)$, where $\Delta < 1$ denotes the probability gap between the correct and second most likely answers. Next, we present an expressiveness result for the self-correction approach with verifier feedback: it enables Transformers to simulate online learning over a pool of experts at test time. Therefore, a single Transformer architecture can provably solve multiple tasks without prior knowledge of the specific task associated with a user query, extending the representation theory of Transformers from single-task to multi-task settings. Finally, we empirically validate our theoretical results, demonstrating the practical effectiveness of self-correction methods.
Authors: Zhuoxuan Cai, Jian Zhang, Xinbin Yuan, Peng-Tao Jiang, Wenxiang Chen, Bowen Tang, Lujian Yao, Qiyuan Wang, Jinwen Chen, Bo Li
Abstract: Recent studies demonstrate that multimodal large language models (MLLMs) can proficiently evaluate visual quality through interpretable assessments. However, existing approaches typically treat quality scoring and reasoning descriptions as separate tasks with disjoint optimization objectives, leading to a trade-off: models adept at quality reasoning descriptions struggle with precise score regression, while score-focused models lack interpretability. This limitation hinders the full potential of MLLMs in visual quality assessment, where accuracy and interpretability should be mutually reinforcing. To address this, we propose a unified two-stage training framework comprising a cold-start stage and a reinforcement learning-based fine-tuning stage. Specifically, in the first stage, we distill high-quality data from a teacher model through expert-designed prompts, initializing reasoning capabilities via cross-entropy loss supervision. In the second stage, we introduce a novel reward with Group Relative Policy Optimization (GRPO) to jointly optimize scoring accuracy and reasoning consistency. We designate the models derived from these two stages as Q-Ponder-CI and Q-Ponder. Extensive experiments show that Q-Ponder achieves state-of-the-art (SOTA) performance on quality score regression benchmarks, delivering up to 6.5% higher SRCC on cross-domain datasets. Furthermore, Q-Ponder significantly outperforms description-based SOTA models, including its teacher model Qwen-2.5-VL-72B, particularly in description accuracy and reasonableness, demonstrating the generalization potential over diverse tasks.
Authors: Sajjad Abdoli, Freeman Lewin, Gediminas Vasiliauskas, Fabian Schonholz
Abstract: The development of modern Artificial Intelligence (AI) models, particularly diffusion-based models employed in computer vision and image generation tasks, is undergoing a paradigmatic shift in development methodologies. Traditionally dominated by a "Model Centric" approach, in which performance gains were primarily pursued through increasingly complex model architectures and hyperparameter optimization, the field is now recognizing a more nuanced "Data-Centric" approach. This emergent framework foregrounds the quality, structure, and relevance of training data as the principal driver of model performance. To operationalize this paradigm shift, we introduce the DataSeeds.AI sample dataset (the "DSD"), initially comprised of approximately 10,610 high-quality human peer-ranked photography images accompanied by extensive multi-tier annotations. The DSD is a foundational computer vision dataset designed to usher in a new standard for commercial image datasets. Representing a small fraction of DataSeeds.AI's 100 million-plus image catalog, the DSD provides a scalable foundation necessary for robust commercial and multimodal AI development. Through this in-depth exploratory analysis, we document the quantitative improvements generated by the DSD on specific models against known benchmarks and make the code and the trained models used in our evaluation publicly available.
Authors: Fardis Nadimi, Payam Abdisarabshali, Kasra Borazjani, Jacob Chakareski, Seyyedali Hosseinalipour
Abstract: Extended reality (XR) systems, which consist of virtual reality (VR), augmented reality (AR), and mixed reality (XR), offer a transformative interface for immersive, multi-modal, and embodied human-computer interaction. In this paper, we envision that multi-modal multi-task (M3T) federated foundation models (FedFMs) can offer transformative capabilities for XR systems through integrating the representational strength of M3T foundation models (FMs) with the privacy-preserving model training principles of federated learning (FL). We present a modular architecture for FedFMs, which entails different coordination paradigms for model training and aggregations. Central to our vision is the codification of XR challenges that affect the implementation of FedFMs under the SHIFT dimensions: (1) Sensor and modality diversity, (2) Hardware heterogeneity and system-level constraints, (3) Interactivity and embodied personalization, (4) Functional/task variability, and (5) Temporality and environmental variability. We illustrate the manifestation of these dimensions across a set of emerging and anticipated applications of XR systems. Finally, we propose evaluation metrics, dataset requirements, and design tradeoffs necessary for the development of resource-aware FedFMs in XR. This perspective aims to chart the technical and conceptual foundations for context-aware privacy-preserving intelligence in the next generation of XR systems.
Authors: Yijia Shao, Humishka Zope, Yucheng Jiang, Jiaxin Pei, David Nguyen, Erik Brynjolfsson, Diyi Yang
Abstract: The rapid rise of compound AI systems (a.k.a., AI agents) is reshaping the labor market, raising concerns about job displacement, diminished human agency, and overreliance on automation. Yet, we lack a systematic understanding of the evolving landscape. In this paper, we address this gap by introducing a novel auditing framework to assess which occupational tasks workers want AI agents to automate or augment, and how those desires align with the current technological capabilities. Our framework features an audio-enhanced mini-interview to capture nuanced worker desires and introduces the Human Agency Scale (HAS) as a shared language to quantify the preferred level of human involvement. Using this framework, we construct the WORKBank database, building on the U.S. Department of Labor's O*NET database, to capture preferences from 1,500 domain workers and capability assessments from AI experts across over 844 tasks spanning 104 occupations. Jointly considering the desire and technological capability divides tasks in WORKBank into four zones: Automation "Green Light" Zone, Automation "Red Light" Zone, R&D Opportunity Zone, Low Priority Zone. This highlights critical mismatches and opportunities for AI agent development. Moving beyond a simple automate-or-not dichotomy, our results reveal diverse HAS profiles across occupations, reflecting heterogeneous expectations for human involvement. Moreover, our study offers early signals of how AI agent integration may reshape the core human competencies, shifting from information-focused skills to interpersonal ones. These findings underscore the importance of aligning AI agent development with human desires and preparing workers for evolving workplace dynamics.
Authors: Yifan Zeng
Abstract: Large Language Models (LLMs) hold immense promise for revolutionizing financial analysis and decision-making, yet their direct application is often hampered by issues of data hallucination and lack of access to real-time, verifiable financial information. This paper introduces QuantMCP, a novel framework designed to rigorously ground LLMs in financial reality. By leveraging the Model Context Protocol (MCP) for standardized and secure tool invocation, QuantMCP enables LLMs to accurately interface with a diverse array of Python-accessible financial data APIs (e.g., Wind, yfinance). Users can interact via natural language to precisely retrieve up-to-date financial data, thereby overcoming LLM's inherent limitations in factual data recall. More critically, once furnished with this verified, structured data, the LLM's analytical capabilities are unlocked, empowering it to perform sophisticated data interpretation, generate insights, and ultimately support more informed financial decision-making processes. QuantMCP provides a robust, extensible, and secure bridge between conversational AI and the complex world of financial data, aiming to enhance both the reliability and the analytical depth of LLM applications in finance.
Authors: Ala Yankouskaya, Areej B. Babiker, Syeda W. F. Rizvi, Sameha Alshakhsi, Magnus Liebherr, Raian Ali
Abstract: There is growing interest in understanding how people interact with large language models (LLMs) and whether such models elicit dependency or even addictive behaviour. Validated tools to assess the extent to which individuals may become dependent on LLMs are scarce and primarily build on classic behavioral addiction symptoms, adapted to the context of LLM use. We view this as a conceptual limitation, as the LLM-human relationship is more nuanced and warrants a fresh and distinct perspective. To address this gap, we developed and validated a new 12-item questionnaire to measure LLM dependency, referred to as LLM-D12. The scale was based on the authors' prior theoretical work, with items developed accordingly and responses collected from 526 participants in the UK. Exploratory and confirmatory factor analyses, performed on separate halves of the total sample using a split-sample approach, supported a two-factor structure: Instrumental Dependency (six items) and Relationship Dependency (six items). Instrumental Dependency reflects the extent to which individuals rely on LLMs to support or collaborate in decision-making and cognitive tasks. Relationship Dependency captures the tendency to perceive LLMs as socially meaningful, sentient, or companion-like entities. The two-factor structure demonstrated excellent internal consistency and clear discriminant validity. External validation confirmed both the conceptual foundation and the distinction between the two subscales. The psychometric properties and structure of our LLM-D12 scale were interpreted in light of the emerging view that dependency on LLMs does not necessarily indicate dysfunction but may still reflect reliance levels that could become problematic in certain contexts.
Authors: Wenke Xia, Yichu Yang, Hongtao Wu, Xiao Ma, Tao Kong, Di Hu
Abstract: Establishing a reliable and iteratively refined robotic system is essential for deploying real-world applications. While Vision-Language-Action (VLA) models are widely recognized as the foundation model for such robotic deployment, their dependence on expert demonstrations hinders the crucial capabilities of correction and learning from failures. To mitigate this limitation, we introduce a Human-assisted Action Preference Optimization method named HAPO, designed to correct deployment failures and foster effective adaptation through preference alignment for VLA models. This method begins with a human-robot collaboration framework for reliable failure correction and interaction trajectory collection through human intervention. These human-intervention trajectories are further employed within the action preference optimization process, facilitating VLA models to mitigate failure action occurrences while enhancing corrective action adaptation. Specifically, we propose an adaptive reweighting algorithm to address the issues of irreversible interactions and token probability mismatch when introducing preference optimization into VLA models, facilitating model learning from binary desirability signals derived from interactions. Through combining these modules, our human-assisted action preference optimization method ensures reliable deployment and effective learning from failure for VLA models. The experiments conducted in simulation and real-world scenarios prove superior generalization and robustness of our framework across a variety of manipulation tasks.
Authors: Jinyoung Park, Jeehye Na, Jinyoung Kim, Hyunwoo J. Kim
Abstract: Recent works have demonstrated the effectiveness of reinforcement learning (RL)-based post-training in enhancing the reasoning capabilities of large language models (LLMs). In particular, Group Relative Policy Optimization (GRPO) has shown impressive success by employing a PPO-style reinforcement algorithm with group-based normalized rewards. However, the application of GRPO to Video Large Language Models (Video LLMs) has been less studied. In this paper, we explore GRPO for video LLMs and identify two primary issues that impede its effective learning: (1) reliance on safeguards, and (2) the vanishing advantage problem. To mitigate these challenges, we propose DeepVideo-R1, a video large language model trained with our proposed Reg-GRPO (Regressive GRPO) and difficulty-aware data augmentation strategy. Reg-GRPO reformulates the GRPO objective as a regression task, directly predicting the advantage in GRPO. This design eliminates the need for safeguards like clipping and min functions, thereby facilitating more direct policy guidance by aligning the model with the advantage values. We also design the difficulty-aware data augmentation strategy that dynamically augments training samples at solvable difficulty levels, fostering diverse and informative reward signals. Our comprehensive experiments show that DeepVideo-R1 significantly improves video reasoning performance across multiple video reasoning benchmarks.
Authors: Mengsong Wu, YaFei Wang, Yidong Ming, Yuqi An, Yuwei Wan, Wenliang Chen, Binbin Lin, Yuqiang Li, Tong Xie, Dongzhan Zhou
Abstract: Large language models (LLMs) have recently demonstrated promising capabilities in chemistry tasks while still facing challenges due to outdated pretraining knowledge and the difficulty of incorporating specialized chemical expertise. To address these issues, we propose an LLM-based agent that synergistically integrates 137 external chemical tools created ranging from basic information retrieval to complex reaction predictions, and a dataset curation pipeline to generate the dataset ChemToolBench that facilitates both effective tool selection and precise parameter filling during fine-tuning and evaluation. We introduce a Hierarchical Evolutionary Monte Carlo Tree Search (HE-MCTS) framework, enabling independent optimization of tool planning and execution. By leveraging self-generated data, our approach supports step-level fine-tuning (FT) of the policy model and training task-adaptive PRM and ORM that surpass GPT-4o. Experimental evaluations demonstrate that our approach significantly improves performance in Chemistry QA and discovery tasks, offering a robust solution to integrate specialized tools with LLMs for advanced chemical applications. All datasets and code are available at https://github.com/AI4Chem/ChemistryAgent .
Authors: Kuo Yang, Xingjie Yang, Linhui Yu, Qing Xu, Yan Fang, Xu Wang, Zhengyang Zhou, Yang Wang
Abstract: Large Language Model (LLM)-driven Multi-agent systems (Mas) have recently emerged as a powerful paradigm for tackling complex real-world tasks. However, existing Mas construction methods typically rely on manually crafted interaction mechanisms or heuristic rules, introducing human biases and constraining the autonomous ability. Even with recent advances in adaptive Mas construction, existing systems largely remain within the paradigm of semi-autonomous patterns. In this work, we propose MasHost, a Reinforcement Learning (RL)-based framework for autonomous and query-adaptive Mas design. By formulating Mas construction as a graph search problem, our proposed MasHost jointly samples agent roles and their interactions through a unified probabilistic sampling mechanism. Beyond the accuracy and efficiency objectives pursued in prior works, we introduce component rationality as an additional and novel design principle in Mas. To achieve this multi-objective optimization, we propose Hierarchical Relative Policy Optimization (HRPO), a novel RL strategy that collaboratively integrates group-relative advantages and action-wise rewards. To our knowledge, our proposed MasHost is the first RL-driven framework for autonomous Mas graph construction. Extensive experiments on six benchmarks demonstrate that MasHost consistently outperforms most competitive baselines, validating its effectiveness, efficiency, and structure rationality.
Authors: Hee Suk Yoon, Eunseop Yoon, Mark Hasegawa-Johnson, Sungwoong Kim, Chang D. Yoo
Abstract: We introduce ConfPO, a method for preference learning in Large Language Models (LLMs) that identifies and optimizes preference-critical tokens based solely on the training policy's confidence, without requiring any auxiliary models or compute. Unlike prior Direct Alignment Algorithms (DAAs) such as Direct Preference Optimization (DPO), which uniformly adjust all token probabilities regardless of their relevance to preference, ConfPO focuses optimization on the most impactful tokens. This targeted approach improves alignment quality while mitigating overoptimization (i.e., reward hacking) by using the KL divergence budget more efficiently. In contrast to recent token-level methods that rely on credit-assignment models or AI annotators, raising concerns about scalability and reliability, ConfPO is simple, lightweight, and model-free. Experimental results on challenging alignment benchmarks, including AlpacaEval 2 and Arena-Hard, demonstrate that ConfPO consistently outperforms uniform DAAs across various LLMs, delivering better alignment with zero additional computational overhead.
Authors: Griffin Pitts, Sanaz Motamedi
Abstract: As AI chatbots become increasingly integrated in education, students are turning to these systems for guidance, feedback, and information. However, the anthropomorphic characteristics of these chatbots create ambiguity regarding whether students develop trust toward them as they would a human peer or instructor, based in interpersonal trust, or as they would any other piece of technology, based in technology trust. This ambiguity presents theoretical challenges, as interpersonal trust models may inappropriately ascribe human intentionality and morality to AI, while technology trust models were developed for non-social technologies, leaving their applicability to anthropomorphic systems unclear. To address this gap, we investigate how human-like and system-like trusting beliefs comparatively influence students' perceived enjoyment, trusting intention, behavioral intention to use, and perceived usefulness of an AI chatbot - factors associated with students' engagement and learning outcomes. Through partial least squares structural equation modeling, we found that human-like and system-like trust significantly influenced student perceptions, with varied effects. Human-like trust more strongly predicted trusting intention, while system-like trust better predicted behavioral intention and perceived usefulness. Both had similar effects on perceived enjoyment. Given the partial explanatory power of each type of trust, we propose that students develop a distinct form of trust with AI chatbots (human-AI trust) that differs from human-human and human-technology models of trust. Our findings highlight the need for new theoretical frameworks specific to human-AI trust and offer practical insights for fostering appropriately calibrated trust, which is critical for the effective adoption and pedagogical impact of AI in education.
Authors: Hao Hu, Xinqi Wang, Simon Shaolei Du
Abstract: We introduce a novel task of clustering trajectories from offline reinforcement learning (RL) datasets, where each cluster center represents the policy that generated its trajectories. By leveraging the connection between the KL-divergence of offline trajectory distributions and a mixture of policy-induced distributions, we formulate a natural clustering objective. To solve this, we propose Policy-Guided K-means (PG-Kmeans) and Centroid-Attracted Autoencoder (CAAE). PG-Kmeans iteratively trains behavior cloning (BC) policies and assigns trajectories based on policy generation probabilities, while CAAE resembles the VQ-VAE framework by guiding the latent representations of trajectories toward the vicinity of specific codebook entries to achieve clustering. Theoretically, we prove the finite-step convergence of PG-Kmeans and identify a key challenge in offline trajectory clustering: the inherent ambiguity of optimal solutions due to policy-induced conflicts, which can result in multiple equally valid but structurally distinct clusterings. Experimentally, we validate our methods on the widely used D4RL dataset and custom GridWorld environments. Our results show that both PG-Kmeans and CAAE effectively partition trajectories into meaningful clusters. They offer a promising framework for policy-based trajectory clustering, with broad applications in offline RL and beyond.
Authors: Bingheng Wu, Jingze Shi, Yifan Wu, Nan Tang, Yuyu Luo
Abstract: Transformers exhibit proficiency in capturing long-range dependencies, whereas State Space Models (SSMs) facilitate linear-time sequence modeling. Notwithstanding their synergistic potential, the integration of these architectures presents a significant challenge, primarily attributable to a fundamental incongruity in their respective positional encoding mechanisms: Transformers rely on explicit Rotary Position Embeddings (RoPE), while SSMs leverage implicit positional representations via convolutions. This divergence often precipitates discontinuities and suboptimal performance. To address this impediment, we propose a unified rotary position embedding (Unified RoPE) methodology, thereby establishing a consistent positional encoding framework for both self-attention and state-space components. Using this Unified RoPE, we introduce TransXSSM, a hybrid architecture that coherently integrates the Transformer and SSM layers under this unified positional encoding scheme. At a 4K sequence length, TransXSSM exhibits training and inference speeds that are 42.3\% and 29.5\% faster, respectively, relative to standard Transformer models. It also delivers higher accuracy: under comparable settings, it surpasses a Transformer baseline by over 4\% on language modeling benchmarks.TransXSSM furthermore scales more effectively: TransXSSM-1.3B gains 7.22\% in average accuracy over its 320M version (versus about 6\% gains for equivalent Transformers or SSMs). Our results show that unified positional encoding resolves positional incompatibility in hybrid models, enabling efficient, high-performance long-context modeling.
Authors: Chengpeng Li, Zhengyang Tang, Ziniu Li, Mingfeng Xue, Keqin Bao, Tian Ding, Ruoyu Sun, Benyou Wang, Xiang Wang, Junyang Lin, Dayiheng Liu
Abstract: Large Reasoning Models (LRMs) like o1 and DeepSeek-R1 have shown remarkable progress in natural language reasoning with long chain-of-thought (CoT), yet they remain inefficient or inaccurate when handling complex mathematical operations. Addressing these limitations through computational tools (e.g., computation libraries and symbolic solvers) is promising, but it introduces a technical challenge: Code Interpreter (CI) brings external knowledge beyond the model's internal text representations, thus the direct combination is not efficient. This paper introduces CoRT, a post-training framework for teaching LRMs to leverage CI effectively and efficiently. As a first step, we address the data scarcity issue by synthesizing code-integrated reasoning data through Hint-Engineering, which strategically inserts different hints at appropriate positions to optimize LRM-CI interaction. We manually create 30 high-quality samples, upon which we post-train models ranging from 1.5B to 32B parameters, with supervised fine-tuning, rejection fine-tuning and reinforcement learning. Our experimental results demonstrate that Hint-Engineering models achieve 4\% and 8\% absolute improvements on DeepSeek-R1-Distill-Qwen-32B and DeepSeek-R1-Distill-Qwen-1.5B respectively, across five challenging mathematical reasoning datasets. Furthermore, Hint-Engineering models use about 30\% fewer tokens for the 32B model and 50\% fewer tokens for the 1.5B model compared with the natural language models. The models and code are available at https://github.com/ChengpengLi1003/CoRT.