Authors: Chenrui Fan, Ming Li, Lichao Sun, Tianyi Zhou
Abstract: We find that the response length of reasoning LLMs, whether trained by reinforcement learning or supervised learning, drastically increases for ill-posed questions with missing premises (MiP), ending up with redundant and ineffective thinking. This newly introduced scenario exacerbates the general overthinking issue to a large extent, which we name as the MiP-Overthinking. Such failures are against the ``test-time scaling law'' but have been widely observed on multiple datasets we curated with MiP, indicating the harm of cheap overthinking and a lack of critical thinking. Surprisingly, LLMs not specifically trained for reasoning exhibit much better performance on the MiP scenario, producing much shorter responses that quickly identify ill-posed queries. This implies a critical flaw of the current training recipe for reasoning LLMs, which does not encourage efficient thinking adequately, leading to the abuse of thinking patterns. To further investigate the reasons behind such failures, we conduct fine-grained analyses of the reasoning length, overthinking patterns, and location of critical thinking on different types of LLMs. Moreover, our extended ablation study reveals that the overthinking is contagious through the distillation of reasoning models' responses. These results improve the understanding of overthinking and shed novel insights into mitigating the problem.
Authors: Umakanta Maharana, Sarthak Verma, Avarna Agarwal, Prakashini Mruthyunjaya, Dwarikanath Mahapatra, Sakir Ahmed, Murari Mandal
Abstract: Large language models (LLMs) offer a promising pre-screening tool, improving early disease detection and providing enhanced healthcare access for underprivileged communities. The early diagnosis of various diseases continues to be a significant challenge in healthcare, primarily due to the nonspecific nature of early symptoms, the shortage of expert medical practitioners, and the need for prolonged clinical evaluations, all of which can delay treatment and adversely affect patient outcomes. With impressive accuracy in prediction across a range of diseases, LLMs have the potential to revolutionize clinical pre-screening and decision-making for various medical conditions. In this work, we study the diagnostic capability of LLMs for Rheumatoid Arthritis (RA) with real world patients data. Patient data was collected alongside diagnoses from medical experts, and the performance of LLMs was evaluated in comparison to expert diagnoses for RA disease prediction. We notice an interesting pattern in disease diagnosis and find an unexpected \textit{misalignment between prediction and explanation}. We conduct a series of multi-round analyses using different LLM agents. The best-performing model accurately predicts rheumatoid arthritis (RA) diseases approximately 95\% of the time. However, when medical experts evaluated the reasoning generated by the model, they found that nearly 68\% of the reasoning was incorrect. This study highlights a clear misalignment between LLMs high prediction accuracy and its flawed reasoning, raising important questions about relying on LLM explanations in clinical settings. \textbf{LLMs provide incorrect reasoning to arrive at the correct answer for RA disease diagnosis.}
Authors: Yuxin Wang, Yiran Guo, Yining Zheng, Zhangyue Yin, Shuo Chen, Jie Yang, Jiajun Chen, Xuanjing Huang, Xipeng Qiu
Abstract: The integration of tool learning with Large Language Models (LLMs) has expanded their capabilities in handling complex tasks by leveraging external tools. However, existing benchmarks for tool learning inadequately address critical real-world personalized scenarios, particularly those requiring multi-hop reasoning and inductive knowledge adaptation in dynamic environments. To bridge this gap, we introduce FamilyTool, a novel benchmark grounded in a family-based knowledge graph (KG) that simulates personalized, multi-hop tool use scenarios. FamilyTool challenges LLMs with queries spanning 1 to 3 relational hops (e.g., inferring familial connections and preferences) and incorporates an inductive KG setting where models must adapt to unseen user preferences and relationships without re-training, a common limitation in prior approaches that compromises generalization. We further propose KGETool: a simple KG-augmented evaluation pipeline to systematically assess LLMs' tool use ability in these settings. Experiments reveal significant performance gaps in state-of-the-art LLMs, with accuracy dropping sharply as hop complexity increases and inductive scenarios exposing severe generalization deficits. These findings underscore the limitations of current LLMs in handling personalized, evolving real-world contexts and highlight the urgent need for advancements in tool-learning frameworks. FamilyTool serves as a critical resource for evaluating and advancing LLM agents' reasoning, adaptability, and scalability in complex, dynamic environments. Code and dataset are available at Github.
Authors: Kostas Hatalis, Despina Christou, Vyshnavi Kondapalli
Abstract: Agents powered by Large Language Models (LLMs) have recently demonstrated impressive capabilities in various tasks. Still, they face limitations in tasks requiring specific, structured knowledge, flexibility, or accountable decision-making. While agents are capable of perceiving their environments, forming inferences, planning, and executing actions towards goals, they often face issues such as hallucinations and lack of contextual memory across interactions. This paper explores how Case-Based Reasoning (CBR), a strategy that solves new problems by referencing past experiences, can be integrated into LLM agent frameworks. This integration allows LLMs to leverage explicit knowledge, enhancing their effectiveness. We systematically review the theoretical foundations of these enhanced agents, identify critical framework components, and formulate a mathematical model for the CBR processes of case retrieval, adaptation, and learning. We also evaluate CBR-enhanced agents against other methods like Chain-of-Thought reasoning and standard Retrieval-Augmented Generation, analyzing their relative strengths. Moreover, we explore how leveraging CBR's cognitive dimensions (including self-reflection, introspection, and curiosity) via goal-driven autonomy mechanisms can further enhance the LLM agent capabilities. Contributing to the ongoing research on neuro-symbolic hybrid systems, this work posits CBR as a viable technique for enhancing the reasoning skills and cognitive aspects of autonomous LLM agents.
Authors: Isma\"il Baaj, Pierre Marquis
Abstract: In this article, we introduce a neuro-symbolic approach that combines a low-level perception task performed by a neural network with a high-level reasoning task performed by a possibilistic rule-based system. The goal is to be able to derive for each input instance the degree of possibility that it belongs to a target (meta-)concept. This (meta-)concept is connected to intermediate concepts by a possibilistic rule-based system. The probability of each intermediate concept for the input instance is inferred using a neural network. The connection between the low-level perception task and the high-level reasoning task lies in the transformation of neural network outputs modeled by probability distributions (through softmax activation) into possibility distributions. The use of intermediate concepts is valuable for the explanation purpose: using the rule-based system, the classification of an input instance as an element of the (meta-)concept can be justified by the fact that intermediate concepts have been recognized. From the technical side, our contribution consists of the design of efficient methods for defining the matrix relation and the equation system associated with a possibilistic rule-based system. The corresponding matrix and equation are key data structures used to perform inferences from a possibilistic rule-based system and to learn the values of the rule parameters in such a system according to a training data sample. Furthermore, leveraging recent results on the handling of inconsistent systems of fuzzy relational equations, an approach for learning rule parameters according to multiple training data samples is presented. Experiments carried out on the MNIST addition problems and the MNIST Sudoku puzzles problems highlight the effectiveness of our approach compared with state-of-the-art neuro-symbolic ones.
Authors: Boyuan Zheng, Michael Y. Fatemi, Xiaolong Jin, Zora Zhiruo Wang, Apurva Gandhi, Yueqi Song, Yu Gu, Jayanth Srinivasa, Gaowen Liu, Graham Neubig, Yu Su
Abstract: To survive and thrive in complex environments, humans have evolved sophisticated self-improvement mechanisms through environment exploration, hierarchical abstraction of experiences into reuseable skills, and collaborative construction of an ever-growing skill repertoire. Despite recent advancements, autonomous web agents still lack crucial self-improvement capabilities, struggling with procedural knowledge abstraction, refining skills, and skill composition. In this work, we introduce SkillWeaver, a skill-centric framework enabling agents to self-improve by autonomously synthesizing reusable skills as APIs. Given a new website, the agent autonomously discovers skills, executes them for practice, and distills practice experiences into robust APIs. Iterative exploration continually expands a library of lightweight, plug-and-play APIs, significantly enhancing the agent's capabilities. Experiments on WebArena and real-world websites demonstrate the efficacy of SkillWeaver, achieving relative success rate improvements of 31.8% and 39.8%, respectively. Additionally, APIs synthesized by strong agents substantially enhance weaker agents through transferable skills, yielding improvements of up to 54.3% on WebArena. These results demonstrate the effectiveness of honing diverse website interactions into APIs, which can be seamlessly shared among various web agents.
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: Xuechen Liang, Meiling Tao, Yinghui Xia, Tianyu Shi, Jun Wang, JingSong Yang
Abstract: Open large language models (LLMs) have significantly advanced the field of natural language processing, showcasing impressive performance across various tasks.Despite the significant advancements in LLMs, their effective operation still relies heavily on human input to accurately guide the dialogue flow, with agent tuning being a crucial optimization technique that involves human adjustments to the model for better response to such guidance.Addressing this dependency, our work introduces the TinyAgent model, trained on a meticulously curated high-quality dataset. We also present the Collaborative Multi-Agent Tuning (CMAT) framework, an innovative system designed to augment language agent capabilities through adaptive weight updates based on environmental feedback. This framework fosters collaborative learning and real-time adaptation among multiple intelligent agents, enhancing their context-awareness and long-term memory. In this research, we propose a new communication agent framework that integrates multi-agent systems with environmental feedback mechanisms, offering a scalable method to explore cooperative behaviors. Notably, our TinyAgent-7B model exhibits performance on par with GPT-3.5, despite having fewer parameters, signifying a substantial improvement in the efficiency and effectiveness of LLMs.
Authors: Runzhen Xue, Hao Wu, Mingyu Yan, Ziheng Xiao, Xiaochun Ye, Dongrui Fan
Abstract: Design space exploration (DSE) enables architects to systematically evaluate various design options, guiding decisions on the most suitable configurations to meet specific objectives such as optimizing performance, power, and area. However, the growing complexity of modern CPUs has dramatically increased the number of micro-architectural parameters and expanded the overall design space, making DSE more challenging and time-consuming. Existing DSE frameworks struggle in large-scale design spaces due to inaccurate models and limited insights into parameter impact, hindering efficient identification of optimal micro-architectures within tight timeframes. In this work, we introduce AttentionDSE. Its key idea is to use the attention mechanism to establish a direct mapping of micro-architectural parameters to their contributions to predicted performance. This approach enhances both the prediction accuracy and interpretability of the performance model. Furthermore, the weights are dynamically adjusted, enabling the model to respond to design changes and effectively pinpoint the key micro-architectural parameters/components responsible for performance bottlenecks. Thus, AttentionDSE accurately, purposefully, and rapidly discovers optimal designs. Experiments on SPEC 2017 demonstrate that AttentionDSE significantly reduces exploration time by over 80\% and achieves 3.9\% improvement in Pareto Hypervolume compared to state-of-the-art DSE frameworks while maintaining superior prediction accuracy and efficiency with an increasing number of parameters.
Authors: Asma Yamani, Malak Baslyman, Moataz Ahmed
Abstract: AI systems are gaining widespread adoption across various sectors and domains. Creating high-quality AI system requirements is crucial for aligning the AI system with business goals and consumer values and for social responsibility. However, with the uncertain nature of AI systems and the heavy reliance on sensitive data, more research is needed to address the elicitation and analysis of AI systems requirements. With the proprietary nature of many AI systems, there is a lack of open-source requirements artifacts and technical requirements documents for AI systems, limiting broader research and investigation. With Large Language Models (LLMs) emerging as a promising alternative to human-generated text, this paper investigates the potential use of LLMs to generate user stories for AI systems based on abstracts from scholarly papers. We conducted an empirical evaluation using three LLMs and generated $1260$ user stories from $42$ abstracts from $26$ domains. We assess their quality using the Quality User Story (QUS) framework. Moreover, we identify relevant non-functional requirements (NFRs) and ethical principles. Our analysis demonstrates that the investigated LLMs can generate user stories inspired by the needs of various stakeholders, offering a promising approach for generating user stories for research purposes and for aiding in the early requirements elicitation phase of AI systems. We have compiled and curated a collection of stories generated by various LLMs into a dataset (UStAI), which is now publicly available for use.
Authors: Yiming Tang, Yi Fan, Chenxiao Yu, Tiankai Yang, Yue Zhao, Xiyang Hu
Abstract: The integration of large language models (LLMs) into information retrieval systems introduces new attack surfaces, particularly for adversarial ranking manipulations. We present StealthRank, a novel adversarial ranking attack that manipulates LLM-driven product recommendation systems while maintaining textual fluency and stealth. Unlike existing methods that often introduce detectable anomalies, StealthRank employs an energy-based optimization framework combined with Langevin dynamics to generate StealthRank Prompts (SRPs)-adversarial text sequences embedded within product descriptions that subtly yet effectively influence LLM ranking mechanisms. We evaluate StealthRank across multiple LLMs, demonstrating its ability to covertly boost the ranking of target products while avoiding explicit manipulation traces that can be easily detected. Our results show that StealthRank consistently outperforms state-of-the-art adversarial ranking baselines in both effectiveness and stealth, highlighting critical vulnerabilities in LLM-driven recommendation systems.
Authors: Yin Wu, Zhengxuan Zhang, Fuling Wang, Yuyu Luo, Hui Xiong, Nan Tang
Abstract: Misinformation continues to pose a significant challenge in today's information ecosystem, profoundly shaping public perception and behavior. Among its various manifestations, Out-of-Context (OOC) misinformation is particularly obscure, as it distorts meaning by pairing authentic images with misleading textual narratives. Existing methods for detecting OOC misinformation predominantly rely on coarse-grained similarity metrics between image-text pairs, which often fail to capture subtle inconsistencies or provide meaningful explainability. While multi-modal large language models (MLLMs) demonstrate remarkable capabilities in visual reasoning and explanation generation, they have not yet demonstrated the capacity to address complex, fine-grained, and cross-modal distinctions necessary for robust OOC detection. To overcome these limitations, we introduce EXCLAIM, a retrieval-based framework designed to leverage external knowledge through multi-granularity index of multi-modal events and entities. Our approach integrates multi-granularity contextual analysis with a multi-agent reasoning architecture to systematically evaluate the consistency and integrity of multi-modal news content. Comprehensive experiments validate the effectiveness and resilience of EXCLAIM, demonstrating its ability to detect OOC misinformation with 4.3% higher accuracy compared to state-of-the-art approaches, while offering explainable and actionable insights.
Authors: Wenqiao Zhu, Lulu Wang, Jun Wu
Abstract: Predicting Click-Through Rates is a crucial function within recommendation and advertising platforms, as the output of CTR prediction determines the order of items shown to users. The Embedding \& MLP paradigm has become a standard approach for industrial recommendation systems and has been widely deployed. However, this paradigm suffers from cold-start problems, where there is either no or only limited user action data available, leading to poorly learned ID embeddings. The cold-start problem hampers the performance of new items. To address this problem, we designed a novel diffusion model to generate a warmed-up embedding for new items. Specifically, we define a novel diffusion process between the ID embedding space and the side information space. In addition, we can derive a sub-sequence from the diffusion steps to expedite training, given that our diffusion model is non-Markovian. Our diffusion model is supervised by both the variational inference and binary cross-entropy objectives, enabling it to generate warmed-up embeddings for items in both the cold-start and warm-up phases. Additionally, we have conducted extensive experiments on three recommendation datasets. The results confirmed the effectiveness of our approach.
Authors: Yikuan Xia, Jiazun Chen, Yirui Zhan, Suifeng Zhao, Weipeng Jiang, Chaorui Zhang, Wei Han, Bo Bai, Jun Gao
Abstract: Large language models (LLMs) excel in question-answering (QA) tasks, and retrieval-augmented generation (RAG) enhances their precision by incorporating external evidence from diverse sources like web pages, databases, and knowledge graphs. However, current RAG methods rely on agent-specific strategies for individual data sources, posing challenges low-resource or black-box environments and complicates operations when evidence is fragmented across sources. To address these limitations, we propose ER-RAG, a framework that unifies evidence integration across heterogeneous data sources using the Entity-Relationship (ER) model. ER-RAG standardizes entity retrieval and relationship querying through ER-based APIs with GET and JOIN operations. It employs a two-stage generation process: first, a preference optimization module selects optimal sources; second, another module constructs API chains based on source schemas. This unified approach allows efficient fine-tuning and seamless integration across diverse data sources. ER-RAG demonstrated its effectiveness by winning all three tracks of the 2024 KDDCup CRAG Challenge, achieving performance on par with commercial RAG pipelines using an 8B LLM backbone. It outperformed hybrid competitors by 3.1% in LLM score and accelerated retrieval by 5.5X.
Authors: Kevin Dela Rosa
Abstract: We present RAVEN an adaptive AI agent framework designed for multimodal entity discovery and retrieval in large-scale video collections. Synthesizing information across visual, audio, and textual modalities, RAVEN autonomously processes video data to produce structured, actionable representations for downstream tasks. Key contributions include (1) a category understanding step to infer video themes and general-purpose entities, (2) a schema generation mechanism that dynamically defines domain-specific entities and attributes, and (3) a rich entity extraction process that leverages semantic retrieval and schema-guided prompting. RAVEN is designed to be model-agnostic, allowing the integration of different vision-language models (VLMs) and large language models (LLMs) based on application-specific requirements. This flexibility supports diverse applications in personalized search, content discovery, and scalable information retrieval, enabling practical applications across vast datasets.
Authors: Jiaming Luo, Weiyi Luo, Guoqing Sun, Mengchen Zhu, Haifeng Tang, Kunyao Lan, Mengyue Wu, Kenny Q. Zhu
Abstract: Designing effective debt collection systems is crucial for improving operational efficiency and reducing costs in the financial industry. However, the challenges of maintaining script diversity, contextual relevance, and coherence make this task particularly difficult. This paper presents a debt collection system based on real debtor-collector data from a major commercial bank. We construct a script library from real-world debt collection conversations, and propose a two-stage retrieval based response system for contextual relevance. Experimental results show that our system improves script diversity, enhances response relevance, and achieves practical deployment efficiency through knowledge distillation. This work offers a scalable and automated solution, providing valuable insights for advancing debt collection practices in real-world applications.
Authors: Ngoc Luyen Le, Marie-H\'el\`ene Abel
Abstract: Group recommender systems aim to generate recommendations that align with the collective preferences of a group, introducing challenges that differ significantly from those in individual recommendation scenarios. This paper presents Joint Group Profiling and Recommendation via Deep Neural Network-based Multi-Task Learning, a framework that unifies group profiling and recommendation tasks within a single model. By jointly learning these tasks, the model develops a deeper understanding of group dynamics, leading to improved recommendation accuracy. The shared representations between the two tasks facilitate the discovery of latent features essential to both, resulting in richer and more informative group embeddings. To further enhance performance, an attention mechanism is integrated to dynamically evaluate the relevance of different group features and item attributes, ensuring the model prioritizes the most impactful information. Experiments and evaluations on real-world datasets demonstrate that our multi-task learning approach consistently outperforms baseline models in terms of accuracy, validating its effectiveness and robustness.
Authors: Tanzir Hossain, Ar-Rafi Islam, Md. Sabbir Hossain, Annajiat Alim Rasel
Abstract: This study presents a cascaded architecture for extractive summarization of multimedia content via audio-to-text alignment. The proposed framework addresses the challenge of extracting key insights from multimedia sources like YouTube videos. It integrates audio-to-text conversion using Microsoft Azure Speech with advanced extractive summarization models, including Whisper, Pegasus, and Facebook BART XSum. The system employs tools such as Pytube, Pydub, and SpeechRecognition for content retrieval, audio extraction, and transcription. Linguistic analysis is enhanced through named entity recognition and semantic role labeling. Evaluation using ROUGE and F1 scores demonstrates that the cascaded architecture outperforms conventional summarization methods, despite challenges like transcription errors. Future improvements may include model fine-tuning and real-time processing. This study contributes to multimedia summarization by improving information retrieval, accessibility, and user experience.
Authors: Chirag Shah, Hideo Joho, Kirandeep Kaur, Preetam Prabhu Srikar Dammu
Abstract: Recent advancements in generative AI have significantly increased interest in personalized agents. With increased personalization, there is also a greater need for being able to trust decision-making and action taking capabilities of these agents. However, the evaluation methods for these agents remain outdated and inadequate, often failing to capture the dynamic and evolving nature of user interactions. In this conceptual article, we argue for a paradigm shift in evaluating personalized and adaptive agents. We propose a comprehensive novel framework that models user personas with unique attributes and preferences. In this framework, agents interact with these simulated users through structured interviews to gather their preferences and offer customized recommendations. These recommendations are then assessed dynamically using simulations driven by Large Language Models (LLMs), enabling an adaptive and iterative evaluation process. Our flexible framework is designed to support a variety of agents and applications, ensuring a comprehensive and versatile evaluation of recommendation strategies that focus on proactive, personalized, and trustworthy aspects.
Authors: Hongbin Liang, Hezhe Qiao, Wei Huang, Qizhou Wang, Mingsheng Shang, Lin Chen
Abstract: Ensuring the safety of vulnerable road users through accurate prediction of pedestrian crossing intention (PCI) plays a crucial role in the context of autonomous and assisted driving. Analyzing the set of observation video frames in ego-view has been widely used in most PCI prediction methods to forecast the cross intent. However, they struggle to capture the critical events related to pedestrian behaviour along the temporal dimension due to the high redundancy of the video frames, which results in the sub-optimal performance of PCI prediction. Our research addresses the challenge by introducing a novel approach called \underline{T}emporal-\underline{c}ontextual Event \underline{L}earning (TCL). The TCL is composed of the Temporal Merging Module (TMM), which aims to manage the redundancy by clustering the observed video frames into multiple key temporal events. Then, the Contextual Attention Block (CAB) is employed to adaptively aggregate multiple event features along with visual and non-visual data. By synthesizing the temporal feature extraction and contextual attention on the key information across the critical events, TCL can learn expressive representation for the PCI prediction. Extensive experiments are carried out on three widely adopted datasets, including PIE, JAAD-beh, and JAAD-all. The results show that TCL substantially surpasses the state-of-the-art methods. Our code can be accessed at https://github.com/dadaguailhb/TCL.
Authors: Ben Degen
Abstract: Formulating research questions is a foundational yet challenging academic skill, one that generative AI systems often oversimplify by offering instant answers at the expense of student reflection. This protocol lays out a study grounded in constructivist learning theory to evaluate a novel AI-based Socratic Tutor, designed to foster cognitive engagement and scaffold research question development in higher education. Anchored in dialogic pedagogy, the tutor engages students through iterative, reflective questioning, aiming to promote System 2 thinking and counteract overreliance on AI-generated outputs. In a quasi-experimental design, approximately 80 German pre-service biology teacher students will be randomly assigned to one of two groups: an AI Socratic Tutor condition and an uninstructed chatbot control. Across multiple cycles, students are expected to formulate research questions based on background texts, with quality assessed through double-blind expert review. The study also examines transfer of skills to novel phenomena and captures student perceptions through mixed-methods analysis, including surveys, interviews and reflective journals. This study aims to advance the understanding of how generative AI can be pedagogically aligned to support, not replace, human cognition and offers design principles for human-AI collaboration in education.
Authors: Dang Nguyen, Chenhao Tan
Abstract: Understanding and mitigating biases is critical for the adoption of large language models (LLMs) in high-stakes decision-making. We introduce Admissions and Hiring, decision tasks with hypothetical applicant profiles where a person's race can be inferred from their name, as simplified test beds for racial bias. We show that Gemma 2B Instruct and LLaMA 3.2 3B Instruct exhibit strong biases. Gemma grants admission to 26% more White than Black applicants, and LLaMA hires 60% more Asian than White applicants. We demonstrate that these biases are resistant to prompt engineering: multiple prompting strategies all fail to promote fairness. In contrast, using distributed alignment search, we can identify "race subspaces" within model activations and intervene on them to debias model decisions. Averaging the representation across all races within the subspaces reduces Gemma's bias by 37-57%. Finally, we examine the generalizability of Gemma's race subspaces, and find limited evidence for generalization, where changing the prompt format can affect the race representation. Our work suggests mechanistic approaches may provide a promising venue for improving the fairness of LLMs, but a universal race representation remains elusive.
Authors: Shiqin Zeng, Haoyun Li, Abhinav Prakash Gahlot, Felix J. Herrmann
Abstract: This study investigates the use of score-based generative models for reservoir simulation, with a focus on reconstructing spatially varying permeability and saturation fields in saline aquifers, inferred from sparse observations at two well locations. By modeling the joint distribution of permeability and saturation derived from high-fidelity reservoir simulations, the proposed neural network is trained to learn the complex spatiotemporal dynamics governing multiphase fluid flow in porous media. During inference, the framework effectively reconstructs both permeability and saturation fields by conditioning on sparse vertical profiles extracted from well log data. This approach introduces a novel methodology for incorporating physical constraints and well log guidance into generative models, significantly enhancing the accuracy and physical plausibility of the reconstructed subsurface states. Furthermore, the framework demonstrates strong generalization capabilities across varying geological scenarios, highlighting its potential for practical deployment in data-scarce reservoir management tasks.
Authors: Polycarp Nalela, Deepthi Rao, Praveen Rao
Abstract: Cancer remains a leading global health challenge and a major cause of mortality. This study leverages machine learning (ML) to predict the survivability of cancer patients with metastatic patterns using the comprehensive MSK-MET dataset, which includes genomic and clinical data from 25,775 patients across 27 cancer types. We evaluated five ML models-XGBoost, Na\"ive Bayes, Decision Tree, Logistic Regression, and Random Fores using hyperparameter tuning and grid search. XGBoost emerged as the best performer with an area under the curve (AUC) of 0.82. To enhance model interpretability, SHapley Additive exPlanations (SHAP) were applied, revealing key predictors such as metastatic site count, tumor mutation burden, fraction of genome altered, and organ-specific metastases. Further survival analysis using Kaplan-Meier curves, Cox Proportional Hazards models, and XGBoost Survival Analysis identified significant predictors of patient outcomes, offering actionable insights for clinicians. These findings could aid in personalized prognosis and treatment planning, ultimately improving patient care.
Authors: Tahniat Khan, Soroor Motie, Sedef Akinli Kocak, Shaina Raza
Abstract: The rapid adoption of large language models (LLMs) has led to significant energy consumption and carbon emissions, posing a critical challenge to the sustainability of generative AI technologies. This paper explores the integration of energy-efficient optimization techniques in the deployment of LLMs to address these environmental concerns. We present a case study and framework that demonstrate how strategic quantization and local inference techniques can substantially lower the carbon footprints of LLMs without compromising their operational effectiveness. Experimental results reveal that these methods can reduce energy consumption and carbon emissions by up to 45\% post quantization, making them particularly suitable for resource-constrained environments. The findings provide actionable insights for achieving sustainability in AI while maintaining high levels of accuracy and responsiveness.
Authors: Haiping Liu, Hongpeng Zhou
Abstract: Rotary Position Embedding (RoPE) is widely adopted in Transformers due to its ability to encode relative positions with high efficiency and extrapolation capability. However, existing RoPE variants lack a unified theoretical foundation, especially in higher dimensions. In this paper, we propose a systematic mathematical framework for RoPE grounded in Lie group and Lie algebra theory. We identify two core properties of RoPE, named relativity and reversibility, and derive general constraints and constructions for valid RoPE in 1D, 2D, and N-dimensional (ND). We prove that RoPE must lie in the basis of a maximal abelian subalgebra (MASA) of the special orthogonal Lie algebra, and show that standard RoPE corresponds to the maximal toral subalgebra. Furthermore, we propose to model inter-dimensional interactions by learning an orthogonal basis transformation. Our framework unifies and explains existing RoPE designs, while enabling principled extensions to new modalities and tasks.
Authors: Peizhi Niu, Yu-Hsiang Wang, Vishal Rana, Chetan Rupakheti, Abhishek Pandey, Olgica Milenkovic
Abstract: We introduce a new graph diffusion model for small molecule generation, \emph{DMol}, which outperforms the state-of-the-art DiGress model in terms of validity by roughly $1.5\%$ across all benchmarking datasets while reducing the number of diffusion steps by at least $10$-fold, and the running time to roughly one half. The performance improvements are a result of a careful change in the objective function and a ``graph noise" scheduling approach which, at each diffusion step, allows one to only change a subset of nodes of varying size in the molecule graph. Another relevant property of the method is that it can be easily combined with junction-tree-like graph representations that arise by compressing a collection of relevant ring structures into supernodes. Unlike classical junction-tree techniques that involve VAEs and require complicated reconstruction steps, compressed DMol directly performs graph diffusion on a graph that compresses only a carefully selected set of frequent carbon rings into supernodes, which results in straightforward sample generation. This compressed DMol method offers additional validity improvements over generic DMol of roughly $2\%$, increases the novelty of the method, and further improves the running time due to reductions in the graph size.
Authors: Yanhao Dong, Yubo Miao, Weinan Li, Xiao Zheng, Chao Wang, Feng Lyu
Abstract: Large Language Models (LLMs) exhibit pronounced memory-bound characteristics during inference due to High Bandwidth Memory (HBM) bandwidth constraints. In this paper, we propose an L2 Cache-oriented asynchronous KV Cache prefetching method to break through the memory bandwidth bottleneck in LLM inference through computation-load overlap. By strategically scheduling idle memory bandwidth during active computation windows, our method proactively prefetches required KV Cache into GPU L2 cache, enabling high-speed L2 cache hits for subsequent accesses and effectively hiding HBM access latency within computational cycles. Extensive experiments on NVIDIA H20 GPUs demonstrate that the proposed method achieves 2.15x improvement in attention kernel efficiency and up to 1.97x end-to-end throughput enhancement, surpassing state-of-the-art baseline FlashAttention-3. Notably, our solution maintains orthogonality to existing optimization techniques and can be integrated with current inference frameworks, providing a scalable latency-hiding solution for next-generation LLM inference engines.
Authors: Michael Somma
Abstract: Cyberattacks on critical infrastructure, particularly water distribution systems, have increased due to rapid digitalization and the integration of IoT devices and industrial control systems (ICS). These cyber-physical systems (CPS) introduce new vulnerabilities, requiring robust and automated intrusion detection systems (IDS) to mitigate potential threats. This study addresses key challenges in anomaly detection by leveraging time correlations in sensor data, integrating physical principles into machine learning models, and optimizing computational efficiency for edge applications. We build upon the concept of temporal differential consistency (TDC) loss to capture the dynamics of the system, ensuring meaningful relationships between dynamic states. Expanding on this foundation, we propose a hybrid autoencoder-based approach, referred to as hybrid TDC-AE, which extends TDC by incorporating both deterministic nodes and conventional statistical nodes. This hybrid structure enables the model to account for non-deterministic processes. Our approach achieves state-of-the-art classification performance while improving time to detect anomalies by 3%, outperforming the BATADAL challenge leader without requiring domain-specific knowledge, making it broadly applicable. Additionally, it maintains the computational efficiency of conventional autoencoders while reducing the number of fully connected layers, resulting in a more sustainable and efficient solution. The method demonstrates how leveraging physics-inspired consistency principles enhances anomaly detection and strengthens the resilience of cyber-physical systems.
Authors: Thijs Willems, Darion Jin Hotan, Jiawen Cheryl Tang, Norakmal Hakim bin Norhashim, King Wang Poon, Zi An Galvyn Goh, Radha Vinod
Abstract: This chapter critiques the dominant reductionist approach in AI and work studies, which isolates tasks and skills as replaceable components. Instead, it advocates for a systemic perspective that emphasizes the interdependence of tasks, roles, and workplace contexts. Two complementary approaches are proposed: an ethnographic, context-rich method that highlights how AI reconfigures work environments and expertise; and a relational task-based analysis that bridges micro-level work descriptions with macro-level labor trends. The authors argue that effective AI impact assessments must go beyond predicting automation rates to include ethical, well-being, and expertise-related questions. Drawing on empirical case studies, they demonstrate how AI reshapes human-technology relations, professional roles, and tacit knowledge practices. The chapter concludes by calling for a human-centric, holistic framework that guides organizational and policy decisions, balancing technological possibilities with social desirability and sustainability of work.
Authors: Bailey J. Eccles, Leon Wong, Blesson Varghese
Abstract: Extensive compute and memory requirements limit the deployment of large language models (LLMs) on any hardware. Compression methods, such as pruning, can reduce model size, which in turn reduces resource requirements. State-of-the-art pruning is based on coarse-grained methods. They are time-consuming and inherently remove critical model parameters, adversely impacting the quality of the pruned model. This paper introduces projection pruning, a novel fine-grained method for pruning LLMs. In addition, LLM projection pruning is enhanced by a new approach we refer to as composite projection pruning - the synergistic combination of unstructured pruning that retains accuracy and structured pruning that reduces model size. We develop Mosaic, a novel system to create and deploy pruned LLMs using composite projection pruning. Mosaic is evaluated using a range of performance and quality metrics on multiple hardware platforms, LLMs, and datasets. Mosaic is 7.19x faster in producing models than existing approaches. Mosaic models achieve up to 84.2% lower perplexity and 31.4% higher accuracy than models obtained from coarse-grained pruning. Up to 67% faster inference and 68% lower GPU memory use is noted for Mosaic models.
Authors: Monika Jotautaite, Mary Phuong, Chatrik Singh Mangat, Maria Angelica Martinez
Abstract: As large language models (LLMs) increasingly integrate into our daily lives, it becomes crucial to understand their implicit biases and moral tendencies. To address this, we introduce a Moral Foundations LLM dataset (MFD-LLM) grounded in Moral Foundations Theory, which conceptualizes human morality through six core foundations. We propose a novel evaluation method that captures the full spectrum of LLMs' revealed moral preferences by answering a range of real-world moral dilemmas. Our findings reveal that state-of-the-art models have remarkably homogeneous value preferences, yet demonstrate a lack of consistency.
Authors: Ronghui Zhang, Wenbin Xing, Mengran Li, Zihan Wang, Junzhou Chen, Xiaolei Ma, Zhiyuan Liu, Zhengbing He
Abstract: Accurate and refined passenger flow prediction is essential for optimizing the collaborative management of multiple collection and distribution modes in large-scale transportation hubs. Traditional methods often focus only on the overall passenger volume, neglecting the interdependence between different modes within the hub. To address this limitation, we propose MM-STFlowNet, a comprehensive multi-mode prediction framework grounded in dynamic spatial-temporal graph modeling. Initially, an integrated temporal feature processing strategy is implemented using signal decomposition and convolution techniques to address data spikes and high volatility. Subsequently, we introduce the Spatial-Temporal Dynamic Graph Convolutional Recurrent Network (STDGCRN) to capture detailed spatial-temporal dependencies across multiple traffic modes, enhanced by an adaptive channel attention mechanism. Finally, the self-attention mechanism is applied to incorporate various external factors, further enhancing prediction accuracy. Experiments on a real-world dataset from Guangzhounan Railway Station in China demonstrate that MM-STFlowNet achieves state-of-the-art performance, particularly during peak periods, providing valuable insight for transportation hub management.
Authors: Azadeh Alavia, Hossein Akhoundib, Fatemeh Kouchmeshkib, Mojtaba Mahmoodianc, Sanduni Jayasinghec, Yongli Rena, Abdolrahman Alavi
Abstract: Geometric Machine Learning (GML) has shown that respecting non-Euclidean geometry in data spaces can significantly improve performance over naive Euclidean assumptions. In parallel, Quantum Machine Learning (QML) has emerged as a promising paradigm that leverages superposition, entanglement, and interference within quantum state manifolds for learning tasks. This paper offers a unifying perspective by casting QML as a specialized yet more expressive branch of GML. We argue that quantum states, whether pure or mixed, reside on curved manifolds (e.g., projective Hilbert spaces or density-operator manifolds), mirroring how covariance matrices inhabit the manifold of symmetric positive definite (SPD) matrices or how image sets occupy Grassmann manifolds. However, QML also benefits from purely quantum properties, such as entanglement-induced curvature, that can yield richer kernel structures and more nuanced data embeddings. We illustrate these ideas with published and newly discussed results, including hybrid classical -quantum pipelines for diabetic foot ulcer classification and structural health monitoring. Despite near-term hardware limitations that constrain purely quantum solutions, hybrid architectures already demonstrate tangible benefits by combining classical manifold-based feature extraction with quantum embeddings. We present a detailed mathematical treatment of the geometrical underpinnings of quantum states, emphasizing parallels to classical Riemannian geometry and manifold-based optimization. Finally, we outline open research challenges and future directions, including Quantum Large Language Models (LLMs), quantum reinforcement learning, and emerging hardware approaches, demonstrating how synergizing GML and QML principles can unlock the next generation of machine intelligence.
Authors: Hicham Talaoubrid, Anissa Mokraoui, Ismail Ben Ayed, Axel Prouvost, Sonimith Hang, Monit Korn, R\'emi Harvey
Abstract: This paper investigates the application of Low-Rank Adaptation (LoRA) to small models for cross-domain few-shot object detection in aerial images. Originally designed for large-scale models, LoRA helps mitigate overfitting, making it a promising approach for resource-constrained settings. We integrate LoRA into DiffusionDet, and evaluate its performance on the DOTA and DIOR datasets. Our results show that LoRA applied after an initial fine-tuning slightly improves performance in low-shot settings (e.g., 1-shot and 5-shot), while full fine-tuning remains more effective in higher-shot configurations. These findings highlight LoRA's potential for efficient adaptation in aerial object detection, encouraging further research into parameter-efficient fine-tuning strategies for few-shot learning. Our code is available here: https://github.com/HichTala/LoRA-DiffusionDet.
Authors: Jonas Torzewski
Abstract: This article presents a method for estimating the dynamic driving states (position, velocity, acceleration and heading) from noisy measurement data. The proposed approach is effective with both complete and partial observations, producing refined trajectory signals with kinematic consistency, ensuring that velocity is the integral of acceleration and position is the integral of velocity. Additionally, the method accounts for the constraint that vehicles can only move in the direction of their orientation. The method is implemented as a configurable python library that also enables trajectory estimation solely based on position data. Regularization is applied to prevent extreme state variations. A key application is enhancing recorded trajectory data for use as reference inputs in machine learning models. At the end, the article presents the results of the method along with a comparison to ground truth data.
Authors: Jiali Cheng, Hadi Amiri
Abstract: Machine Unlearning aims to remove undesired information from trained models without requiring full retraining from scratch. Despite recent advancements, their underlying loss landscapes and optimization dynamics received less attention. In this paper, we investigate and analyze machine unlearning through the lens of mode connectivity - the phenomenon where independently trained models can be connected by smooth low-loss paths in the parameter space. We define and study mode connectivity in unlearning across a range of overlooked conditions, including connections between different unlearning methods, models trained with and without curriculum learning, and models optimized with first-order and secondorder techniques. Our findings show distinct patterns of fluctuation of different evaluation metrics along the curve, as well as the mechanistic (dis)similarity between unlearning methods. To the best of our knowledge, this is the first study on mode connectivity in the context of machine unlearning.
Authors: Michael K\"olle, Tom Bintener, Maximilian Zorn, Gerhard Stenzel, Leo S\"unkel, Thomas Gabor, Claudia Linnhoff-Popien
Abstract: Quantum computing leverages the unique properties of qubits and quantum parallelism to solve problems intractable for classical systems, offering unparalleled computational potential. However, the optimization of quantum circuits remains critical, especially for noisy intermediate-scale quantum (NISQ) devices with limited qubits and high error rates. Genetic algorithms (GAs) provide a promising approach for efficient quantum circuit synthesis by automating optimization tasks. This work examines the impact of various mutation strategies within a GA framework for quantum circuit synthesis. By analyzing how different mutations transform circuits, it identifies strategies that enhance efficiency and performance. Experiments utilized a fitness function emphasizing fidelity, while accounting for circuit depth and T operations, to optimize circuits with four to six qubits. Comprehensive hyperparameter testing revealed that combining delete and swap strategies outperformed other approaches, demonstrating their effectiveness in developing robust GA-based quantum circuit optimizers.
Authors: Dogus Yuksel, Mehmet Cem Catalbas, Bora Oc
Abstract: As leading examples of large language models, ChatGPT and Gemini claim to provide accurate and unbiased information, emphasizing their commitment to political neutrality and avoidance of personal bias. This research investigates the political tendency of large language models and the existence of differentiation according to the query language. For this purpose, ChatGPT and Gemini were subjected to a political axis test using 14 different languages. The findings of the study suggest that these large language models do exhibit political tendencies, with both models demonstrating liberal and leftist biases. A comparative analysis revealed that Gemini exhibited a more pronounced liberal and left-wing tendency compared to ChatGPT. The study also found that these political biases varied depending on the language used for inquiry. The study delves into the factors that constitute political tendencies and linguistic differentiation, exploring differences in the sources and scope of educational data, structural and grammatical features of languages, cultural and political contexts, and the model's response to linguistic features. From this standpoint, and an ethical perspective, it is proposed that artificial intelligence tools should refrain from asserting a lack of political tendencies and neutrality, instead striving for political neutrality and executing user queries by incorporating these tendencies.
Authors: Yuehan Qin, Shawn Li, Yi Nian, Xinyan Velocity Yu, Yue Zhao, Xuezhe Ma
Abstract: Large language models (LLMs) have shown substantial capacity for generating fluent, contextually appropriate responses. However, they can produce hallucinated outputs, especially when a user query includes one or more false premises-claims that contradict established facts. Such premises can mislead LLMs into offering fabricated or misleading details. Existing approaches include pretraining, fine-tuning, and inference-time techniques that often rely on access to logits or address hallucinations after they occur. These methods tend to be computationally expensive, require extensive training data, or lack proactive mechanisms to prevent hallucination before generation, limiting their efficiency in real-time applications. We propose a retrieval-based framework that identifies and addresses false premises before generation. Our method first transforms a user's query into a logical representation, then applies retrieval-augmented generation (RAG) to assess the validity of each premise using factual sources. Finally, we incorporate the verification results into the LLM's prompt to maintain factual consistency in the final output. Experiments show that this approach effectively reduces hallucinations, improves factual accuracy, and does not require access to model logits or large-scale fine-tuning.
Authors: Fay Elhassan, Niccol\`o Ajroldi, Antonio Orvieto, Jonas Geiping
Abstract: The indistinguishability of AI-generated content from human text raises challenges in transparency and accountability. While several methods exist to watermark models behind APIs, embedding watermark strategies directly into model weights that are later reflected in the outputs of the model is challenging. In this study we propose a strategy to finetune a pair of low-rank adapters of a model, one serving as the text-generating model, and the other as the detector, so that a subtle watermark is embedded into the text generated by the first model and simultaneously optimized for detectability by the second. In this way, the watermarking strategy is fully learned end-to-end. This process imposes an optimization challenge, as balancing watermark robustness, naturalness, and task performance requires trade-offs. We discuss strategies on how to optimize this min-max objective and present results showing the effect of this modification to instruction finetuning.
Authors: Xinyuan Huang, Jiechao Gao
Abstract: Federated Learning (FL) often struggles with data heterogeneity due to the naturally uneven distribution of user data across devices. Federated Neural Architecture Search (NAS) enables collaborative search for optimal model architectures tailored to heterogeneous data to achieve higher accuracy. However, this process is time-consuming due to extensive search space and retraining. To overcome this, we introduce FedMetaNAS, a framework that integrates meta-learning with NAS within the FL context to expedite the architecture search by pruning the search space and eliminating the retraining stage. Our approach first utilizes the Gumbel-Softmax reparameterization to facilitate relaxation of the mixed operations in the search space. We then refine the local search process by incorporating Model-Agnostic Meta-Learning, where a task-specific learner adapts both weights and architecture parameters (alphas) for individual tasks, while a meta learner adjusts the overall model weights and alphas based on the gradient information from task learners. Following the meta-update, we propose soft pruning using the same trick on search space to gradually sparsify the architecture, ensuring that the performance of the chosen architecture remains robust after pruning which allows for immediate use of the model without retraining. Experimental evaluations demonstrate that FedMetaNAS significantly accelerates the search process by more than 50\% with higher accuracy compared to FedNAS.
Authors: Halid Abdulrahim Kadi, Kasim Terzi\'c
Abstract: Robotic research is inherently challenging, requiring expertise in diverse environments and control algorithms. Adapting algorithms to new environments often poses significant difficulties, compounded by the need for extensive hyper-parameter tuning in data-driven methods. To address these challenges, we present Agent-Arena, a Python framework designed to streamline the integration, replication, development, and testing of decision-making policies across a wide range of benchmark environments. Unlike existing frameworks, Agent-Arena is uniquely generalised to support all types of control algorithms and is adaptable to both simulation and real-robot scenarios. Please see our GitHub repository https://github.com/halid1020/agent-arena-v0.
Authors: Sergio Lopez-Caceres, Daniel Santiago-Gonzalez
Abstract: Beams of radioactive heavy ions allow researchers to study rare and unstable atomic nuclei, shedding light into the internal structure of exotic nuclei and on how chemical elements are formed in stars. However, the extraction and transport of radioactive beams rely on time-consuming expert-driven tuning methods, where hundreds of parameters are manually optimized. Here, we introduce a system that uses Artificial Intelligence (AI) to assist in the radioactive beam transport process. We apply our methodology to real-life scenarios showing advantages when compared with standard tuning methods. Our method can be extended to other radioactive beam facilities around the world to improve operational efficiency and enhance scientific output.
Authors: Mingchen Li, Di Zhuang, Keyu Chen, Dumindu Samaraweera, Morris Chang
Abstract: Link prediction in graph data utilizes various algorithms and machine learning/deep learning models to predict potential relationships between graph nodes. This technique has found widespread use in numerous real-world applications, including recommendation systems, community networks, and biological structures. However, recent research has highlighted the vulnerability of link prediction models to adversarial attacks, such as poisoning and evasion attacks. Addressing the vulnerability of these models is crucial to ensure stable and robust performance in link prediction applications. While many works have focused on enhancing the robustness of the Graph Convolution Network (GCN) model, the Variational Graph Auto-Encoder (VGAE), a sophisticated model for link prediction, has not been thoroughly investigated in the context of graph adversarial attacks. To bridge this gap, this article proposes an unweighted graph poisoning attack approach using meta-learning techniques to undermine VGAE's link prediction performance. We conducted comprehensive experiments on diverse datasets to evaluate the proposed method and its parameters, comparing it with existing approaches in similar settings. Our results demonstrate that our approach significantly diminishes link prediction performance and outperforms other state-of-the-art methods.
Authors: Minati Rath, Hema Date
Abstract: This study explores the intersection of continuous-variable quantum computing (CVQC) and classical machine learning, focusing on CVQC data encoding techniques, including Displacement encoding and squeezing encoding, alongside Instantaneous Quantum Polynomial (IQP) encoding from discrete quantum computing. We perform an extensive empirical analysis to assess the impact of these encoding methods on classical machine learning models, such as Logistic Regression, Support Vector Machines, K-Nearest Neighbors, and ensemble methods like Random Forest and LightGBM. Our findings indicate that CVQC-based encoding methods significantly enhance feature expressivity, resulting in improved classification accuracy and F1 scores, especially in high-dimensional and complex datasets. However, these improvements come with varying computational costs, which depend on the complexity of the encoding and the architecture of the machine learning models. Additionally, we examine the trade-off between quantum expressibility and classical learnability, offering valuable insights into the practical feasibility of incorporating these quantum encodings into real-world applications. This study contributes to the growing body of research on quantum-classical hybrid learning, emphasizing the role of CVQC in advancing quantum data representation and its integration into classical machine learning workflows.
Authors: Yu Liu, Sergei V. Kalinin
Abstract: Automated experimentation has the potential to revolutionize scientific discovery, but its effectiveness depends on well-defined optimization targets, which are often uncertain or probabilistic in real-world settings. In this work, we demonstrate the application of Multi-Objective Bayesian Optimization (MOBO) to balance multiple, competing rewards in autonomous experimentation. Using scanning probe microscopy (SPM) imaging, one of the most widely used and foundational SPM modes, we show that MOBO can optimize imaging parameters to enhance measurement quality, reproducibility, and efficiency. A key advantage of this approach is the ability to compute and analyze the Pareto front, which not only guides optimization but also provides physical insights into the trade-offs between different objectives. Additionally, MOBO offers a natural framework for human-in-the-loop decision-making, enabling researchers to fine-tune experimental trade-offs based on domain expertise. By standardizing high-quality, reproducible measurements and integrating human input into AI-driven optimization, this work highlights MOBO as a powerful tool for advancing autonomous scientific discovery.
Authors: Xinyu Liu, Xiaoguang Lin, Xiang Liu, Yong Yang, Hongqian Wang, Qilong Sun
Abstract: Recording the open surgery process is essential for educational and medical evaluation purposes; however, traditional single-camera methods often face challenges such as occlusions caused by the surgeon's head and body, as well as limitations due to fixed camera angles, which reduce comprehensibility of the video content. This study addresses these limitations by employing a multi-viewpoint camera recording system, capturing the surgical procedure from six different angles to mitigate occlusions. We propose a fully supervised learning-based time series prediction method to choose the best shot sequences from multiple simultaneously recorded video streams, ensuring optimal viewpoints at each moment. Our time series prediction model forecasts future camera selections by extracting and fusing visual and semantic features from surgical videos using pre-trained models. These features are processed by a temporal prediction network with TimeBlocks to capture sequential dependencies. A linear embedding layer reduces dimensionality, and a Softmax classifier selects the optimal camera view based on the highest probability. In our experiments, we created five groups of open thyroidectomy videos, each with simultaneous recordings from six different angles. The results demonstrate that our method achieves competitive accuracy compared to traditional supervised methods, even when predicting over longer time horizons. Furthermore, our approach outperforms state-of-the-art time series prediction techniques on our dataset. This manuscript makes a unique contribution by presenting an innovative framework that advances surgical video analysis techniques, with significant implications for improving surgical education and patient safety.
Authors: Amir Yazdanbakhsh
Abstract: For decades, Moore's Law has served as a steadfast pillar in computer architecture and system design, promoting a clear abstraction between hardware and software. This traditional Moore's computing paradigm has deepened the rift between the two, enabling software developers to achieve near-exponential performance gains often without needing to delve deeply into hardware-specific optimizations. Yet today, Moore's Law -- with its once relentless performance gains now diminished to incremental improvements -- faces inevitable physical barriers. This stagnation necessitates a reevaluation of the conventional system design philosophy. The traditional decoupled system design philosophy, which maintains strict abstractions between hardware and software, is increasingly obsolete. The once-clear boundary between software and hardware is rapidly dissolving, replaced by co-design. It is imperative for the computing community to intensify its commitment to hardware-software co-design, elevating system abstractions to first-class citizens and reimagining design principles to satisfy the insatiable appetite of modern computing. Hardware-software co-design is not a recent innovation. To illustrate its historical evolution, I classify its development into five relatively distinct ``epochs''. This post also highlights the growing influence of the architecture community in interdisciplinary teams -- particularly alongside ML researchers -- and explores why current co-design paradigms are struggling in today's computing landscape. Additionally, I will examine the concept of the ``hardware lottery'' and explore directions to mitigate its constraining influence on the next era of computing innovation.
Authors: Hailong Shu, Weiwei Song, Yue Wang, Jiping Zhang
Abstract: Wind direction forecasting plays a crucial role in optimizing wind energy production, but faces significant challenges due to the circular nature of directional data, error accumulation in multi-step forecasting, and complex meteorological interactions. This paper presents a novel model, WaveHiTS, which integrates wavelet transform with Neural Hierarchical Interpolation for Time Series to address these challenges. Our approach decomposes wind direction into U-V components, applies wavelet transform to capture multi-scale frequency patterns, and utilizes a hierarchical structure to model temporal dependencies at multiple scales, effectively mitigating error propagation. Experiments conducted on real-world meteorological data from Inner Mongolia, China demonstrate that WaveHiTS significantly outperforms deep learning models (RNN, LSTM, GRU), transformer-based approaches (TFT, Informer, iTransformer), and hybrid models (EMD-LSTM). The proposed model achieves RMSE values of approximately 19.2{\deg}-19.4{\deg} compared to 56{\deg}-64{\deg} for deep learning recurrent models, maintaining consistent accuracy across all forecasting steps up to 60 minutes ahead. Moreover, WaveHiTS demonstrates superior robustness with vector correlation coefficients (VCC) of 0.985-0.987 and hit rates of 88.5%-90.1%, substantially outperforming baseline models. Ablation studies confirm that each component-wavelet transform, hierarchical structure, and U-V decomposition-contributes meaningfully to overall performance. These improvements in wind direction nowcasting have significant implications for enhancing wind turbine yaw control efficiency and grid integration of wind energy.
Authors: Zhouyang Liu, Ning Liu, Yixin Chen, Jiezhong He, Dongsheng Li
Abstract: Graph Edit Distance (GED) is an important similarity measure in graph retrieval, which quantifies the minimum cost of transforming one graph into another through edit operations, and offers flexibility by allowing customizable operation costs. Recent learning-based approaches approximate GEDs with the distances between representations in vector spaces. However, these methods often struggle with varying operation costs due to neglecting the impact of these costs on determining optimal graph mappings. Furthermore, they rely on isolated node distances as guidance, necessitating inefficient reactive refinements of mappings. To address these issues, we propose Graph Edit Network (GEN), a novel learning-based approach for flexible GED computation. By identifying the limitations of existing methods in capturing flexibility of GED, we introduce a principled yet simple solution that incorporates the operation costs before establishing mappings. To improve matching efficiency, we propose a strategy that proactively optimizes guidance from a graph perspective. This strategy initializes guidance as each node's alignment difficulty and captures the interdependencies between matches within and across graphs through a difficulty propagation mechanism, enabling more informed decisions. As a result, GEN selects optimal matches in a single step, minimizing the need for costly refinements. Results on real-world and synthetic datasets demonstrate the effectiveness, time efficiency, and adaptability of GEN, achieving up to 37.8\% error reduction and 72.7\% inference time reduction compared with state-of-the-art models, while performing robustly under varying cost settings and graph sizes.
Authors: Happy Buzaaba, Alexander Wettig, David Ifeoluwa Adelani, Christiane Fellbaum
Abstract: Large language models (LLMs) have achieved impressive results in a wide range of natural language applications. However, they often struggle to recognize low-resource languages, in particular African languages, which are not well represented in large training corpora. In this paper, we consider how to adapt LLMs to low-resource African languages. We find that combining curated data from African languages with high-quality English educational texts results in a training mix that substantially improves the model's performance on these languages. On the challenging IrokoBench dataset, our models consistently achieve the best performance amongst similarly sized baselines, particularly on knowledge-intensive multiple-choice questions (AfriMMLU). Additionally, on the cross-lingual question answering benchmark AfriQA, our models outperform the base model by over 10%. To better understand the role of English data during training, we translate a subset of 200M tokens into Swahili language and perform an analysis which reveals that the content of these data is primarily responsible for the strong performance. We release our models and data to encourage future research on African languages.
Authors: Daniel Tcheurekdjian, Joshua Klasmeier, Tom Cooney, Christopher McCann, Tyler Fenstermaker
Abstract: We present OPAL (Operant Physical Agent with Language), a novel vision-language-action architecture that introduces topological constraints to flow matching for robotic control. To do so, we further introduce topological attention. Our approach models action sequences as topologically-structured representations with non-trivial constraints. Experimental results across 10 complex manipulation tasks demonstrate OPAL's superior performance compared to previous approaches, including Octo, OpenVLA, and ${\pi}$0. Our architecture achieves significant improvements in zero-shot performance without requiring task-specific fine-tuning, while reducing inference computational requirements by 42%. The theoretical guarantees provided by our topological approach result in more coherent long-horizon action sequences. Our results highlight the potential of constraining the search space of learning problems in robotics by deriving from fundamental physical laws, and the possibility of using topological attention to embed causal understanding into transformer architectures.
Authors: Pinhan Zhao, Yuepeng Wang, Xinyu Wang
Abstract: We present a novel symbolic reasoning engine for SQL which can efficiently generate an input $I$ for $n$ queries $P_1, \cdots, P_n$, such that their outputs on $I$ satisfy a given property (expressed in SMT). This is useful in different contexts, such as disproving equivalence of two SQL queries and disambiguating a set of queries. Our first idea is to reason about an under-approximation of each $P_i$ -- that is, a subset of $P_i$'s input-output behaviors. While it makes our approach both semantics-aware and lightweight, this idea alone is incomplete (as a fixed under-approximation might miss some behaviors of interest). Therefore, our second idea is to perform search over an expressive family of under-approximations (which collectively cover all program behaviors of interest), thereby making our approach complete. We have implemented these ideas in a tool, Polygon, and evaluated it on over 30,000 benchmarks across two tasks (namely, SQL equivalence refutation and query disambiguation). Our evaluation results show that Polygon significantly outperforms all prior techniques.
Authors: Judy Hanwen Shen, Carlos Guestrin
Abstract: Foundation models that are capable of automating cognitive tasks represent a pivotal technological shift, yet their societal implications remain unclear. These systems promise exciting advances, yet they also risk flooding our information ecosystem with formulaic, homogeneous, and potentially misleading synthetic content. Developing benchmarks grounded in real use cases where these risks are most significant is therefore critical. Through a thematic analysis using 2 million language model user prompts, we identify creative composition tasks as a prevalent usage category where users seek help with personal tasks that require everyday creativity. Our fine-grained analysis identifies mismatches between current benchmarks and usage patterns among these tasks. Crucially, we argue that the same use cases that currently lack thorough evaluations can lead to negative downstream impacts. This position paper argues that benchmarks focused on creative composition tasks is a necessary step towards understanding the societal harms of AI-generated content. We call for greater transparency in usage patterns to inform the development of new benchmarks that can effectively measure both the progress and the impacts of models with creative capabilities.
Authors: Rahul Singh Maharjan, Marta Romeo, Angelo Cangelosi
Abstract: Visual emotion analysis or recognition has gained considerable attention due to the growing interest in understanding how images can convey rich semantics and evoke emotions in human perception. However, visual emotion analysis poses distinctive challenges compared to traditional vision tasks, especially due to the intricate relationship between general visual features and the different affective states they evoke, known as the affective gap. Researchers have used deep representation learning methods to address this challenge of extracting generalized features from entire images. However, most existing methods overlook the importance of specific emotional attributes such as brightness, colorfulness, scene understanding, and facial expressions. Through this paper, we introduce A4Net, a deep representation network to bridge the affective gap by leveraging four key attributes: brightness (Attribute 1), colorfulness (Attribute 2), scene context (Attribute 3), and facial expressions (Attribute 4). By fusing and jointly training all aspects of attribute recognition and visual emotion analysis, A4Net aims to provide a better insight into emotional content in images. Experimental results show the effectiveness of A4Net, showcasing competitive performance compared to state-of-the-art methods across diverse visual emotion datasets. Furthermore, visualizations of activation maps generated by A4Net offer insights into its ability to generalize across different visual emotion datasets.
Authors: Joochan Kim, Minjoon Jung, Byoung-Tak Zhang
Abstract: Action recognition models have achieved promising results in understanding instructional videos. However, they often rely on dominant, dataset-specific action sequences rather than true video comprehension, a problem that we define as ordinal bias. To address this issue, we propose two effective video manipulation methods: Action Masking, which masks frames of frequently co-occurring actions, and Sequence Shuffling, which randomizes the order of action segments. Through comprehensive experiments, we demonstrate that current models exhibit significant performance drops when confronted with nonstandard action sequences, underscoring their vulnerability to ordinal bias. Our findings emphasize the importance of rethinking evaluation strategies and developing models capable of generalizing beyond fixed action patterns in diverse instructional videos.
Authors: William De Michele, Abel Armas Cervantes, Lea Frermann
Abstract: Business processes are fundamental to organizational operations, yet their optimization remains challenging due to the timeconsuming nature of manual process analysis. Our paper harnesses Large Language Models (LLMs) to automate value-added analysis, a qualitative process analysis technique that aims to identify steps in the process that do not deliver value. To date, this technique is predominantly manual, time-consuming, and subjective. Our method offers a more principled approach which operates in two phases: first, decomposing high-level activities into detailed steps to enable granular analysis, and second, performing a value-added analysis to classify each step according to Lean principles. This approach enables systematic identification of waste while maintaining the semantic understanding necessary for qualitative analysis. We develop our approach using 50 business process models, for which we collect and publish manual ground-truth labels. Our evaluation, comparing zero-shot baselines with more structured prompts reveals (a) a consistent benefit of structured prompting and (b) promising performance for both tasks. We discuss the potential for LLMs to augment human expertise in qualitative process analysis while reducing the time and subjectivity inherent in manual approaches.
Authors: Sujay Khandagale, Bhawna Juneja, Prabhat Agarwal, Aditya Subramanian, Jaewon Yang, Yuting Wang
Abstract: Modern search systems use a multi-stage architecture to deliver personalized results efficiently. Key stages include retrieval, pre-ranking, full ranking, and blending, which refine billions of items to top selections. The pre-ranking stage, vital for scoring and filtering hundreds of thousands of items down to a few thousand, typically relies on two tower models due to their computational efficiency, despite often lacking in capturing complex interactions. While query-item cross interaction features are paramount for full ranking, integrating them into pre-ranking models presents efficiency-related challenges. In this paper, we introduce InteractRank, a novel two tower pre-ranking model with robust cross interaction features used at Pinterest. By incorporating historical user engagement-based query-item interactions in the scoring function along with the two tower dot product, InteractRank significantly boosts pre-ranking performance with minimal latency and computation costs. In real-world A/B experiments at Pinterest, InteractRank improves the online engagement metric by 6.5% over a BM25 baseline and by 3.7% over a vanilla two tower baseline. We also highlight other components of InteractRank, like real-time user-sequence modeling, and analyze their contributions through offline ablation studies. The code for InteractRank is available at https://github.com/pinterest/atg-research/tree/main/InteractRank.
URLs: https://github.com/pinterest/atg-research/tree/main/InteractRank.
Authors: Chrisantha Fernando, Dylan Banarse, Simon Osindero
Abstract: This paper explores an intrinsic motivation for mutual awareness, hypothesizing that humans possess a fundamental drive to understand \textit{and to be understood} even in the absence of extrinsic rewards. Through simulations of the perceptual crossing paradigm, we explore the effect of various internal reward functions in reinforcement learning agents. The drive to understand is implemented as an active inference type artificial curiosity reward, whereas the drive to be understood is implemented through intrinsic rewards for imitation, influence/impressionability, and sub-reaction time anticipation of the other. Results indicate that while artificial curiosity alone does not lead to a preference for social interaction, rewards emphasizing reciprocal understanding successfully drive agents to prioritize interaction. We demonstrate that this intrinsic motivation can facilitate cooperation in tasks where only one agent receives extrinsic reward for the behaviour of the other.
Authors: Tiange Huang, Yongjun Li
Abstract: Unsupervised multivariate time series anomaly detection (UMTSAD) plays a critical role in various domains, including finance, networks, and sensor systems. In recent years, due to the outstanding performance of deep learning in general sequential tasks, many models have been specialized for deep UMTSAD tasks and have achieved impressive results, particularly those based on the Transformer and self-attention mechanisms. However, the sequence anomaly association assumptions underlying these models are often limited to specific predefined patterns and scenarios, such as concentrated or peak anomaly patterns. These limitations hinder their ability to generalize to diverse anomaly situations, especially where the lack of labels poses significant challenges. To address these issues, we propose AMAD, which integrates \textbf{A}uto\textbf{M}asked Attention for UMTS\textbf{AD} scenarios. AMAD introduces a novel structure based on the AutoMask mechanism and an attention mixup module, forming a simple yet generalized anomaly association representation framework. This framework is further enhanced by a Max-Min training strategy and a Local-Global contrastive learning approach. By combining multi-scale feature extraction with automatic relative association modeling, AMAD provides a robust and adaptable solution to UMTSAD challenges. Extensive experimental results demonstrate that the proposed model achieving competitive performance results compared to SOTA benchmarks across a variety of datasets.
Authors: Songwei Zhao, Yuan Jiang, Zijing Zhang, Yang Yu, Hechang Chen
Abstract: Graph neural networks (GNNs) have shown significant success in learning graph representations. However, recent studies reveal that GNNs often fail to outperform simple MLPs on heterophilous graph tasks, where connected nodes may differ in features or labels, challenging the homophily assumption. Existing methods addressing this issue often overlook the importance of information granularity and rarely consider implicit relationships between distant nodes. To overcome these limitations, we propose the Granular and Implicit Graph Network (GRAIN), a novel GNN model specifically designed for heterophilous graphs. GRAIN enhances node embeddings by aggregating multi-view information at various granularity levels and incorporating implicit data from distant, non-neighboring nodes. This approach effectively integrates local and global information, resulting in smoother, more accurate node representations. We also introduce an adaptive graph information aggregator that efficiently combines multi-granularity and implicit data, significantly improving node representation quality, as shown by experiments on 13 datasets covering varying homophily and heterophily. GRAIN consistently outperforms 12 state-of-the-art models, excelling on both homophilous and heterophilous graphs.
Authors: Xiaohua Feng, Yuyuan Li, Chengye Wang, Junlin Liu, Li Zhang, Chaochao Chen
Abstract: Driven by privacy protection laws and regulations, unlearning in Large Language Models (LLMs) is gaining increasing attention. However, current research often neglects the interpretability of the unlearning process, particularly concerning sample-level unlearning difficulty. Existing studies typically assume a uniform unlearning difficulty across samples. This simplification risks attributing the performance of unlearning algorithms to sample selection rather than the algorithm's design, potentially steering the development of LLM unlearning in the wrong direction. Thus, we investigate the relationship between LLM unlearning and sample characteristics, with a focus on unlearning difficulty. Drawing inspiration from neuroscience, we propose a Memory Removal Difficulty ($\mathrm{MRD}$) metric to quantify sample-level unlearning difficulty. Using $\mathrm{MRD}$, we analyze the characteristics of hard-to-unlearn versus easy-to-unlearn samples. Furthermore, we propose an $\mathrm{MRD}$-based weighted sampling method to optimize existing unlearning algorithms, which prioritizes easily forgettable samples, thereby improving unlearning efficiency and effectiveness. We validate the proposed metric and method using public benchmarks and datasets, with results confirming its effectiveness.
Authors: Xiaohua Feng, Yuyuan Li, Huwei Ji, Jiaming Zhang, Li Zhang, Tianyu Du, Chaochao Chen
Abstract: Despite advances in Preference Alignment (PA) for Large Language Models (LLMs), mainstream methods like Reinforcement Learning with Human Feedback (RLHF) face notable challenges. These approaches require high-quality datasets of positive preference examples, which are costly to obtain and computationally intensive due to training instability, limiting their use in low-resource scenarios. LLM unlearning technique presents a promising alternative, by directly removing the influence of negative examples. However, current research has primarily focused on empirical validation, lacking systematic quantitative analysis. To bridge this gap, we propose a framework to explore the relationship between PA and LLM unlearning. Specifically, we introduce a bi-level optimization-based method to quantify the impact of unlearning specific negative examples on PA performance. Our analysis reveals that not all negative examples contribute equally to alignment improvement when unlearned, and the effect varies significantly across examples. Building on this insight, we pose a crucial question: how can we optimally select and weight negative examples for unlearning to maximize PA performance? To answer this, we propose a framework called Unlearning to Align (U2A), which leverages bi-level optimization to efficiently select and unlearn examples for optimal PA performance. We validate the proposed method through extensive experiments, with results confirming its effectiveness.
Authors: Heather Lent, Erick Galinkin, Yiyi Chen, Jens Myrup Pedersen, Leon Derczynski, Johannes Bjerva
Abstract: As NLP models are used by a growing number of end-users, an area of increasing importance is NLP Security (NLPSec): assessing the vulnerability of models to malicious attacks and developing comprehensive countermeasures against them. While work at the intersection of NLP and cybersecurity has the potential to create safer NLP for all, accidental oversights can result in tangible harm (e.g., breaches of privacy or proliferation of malicious models). In this emerging field, however, the research ethics of NLP have not yet faced many of the long-standing conundrums pertinent to cybersecurity, until now. We thus examine contemporary works across NLPSec, and explore their engagement with cybersecurity's ethical norms. We identify trends across the literature, ultimately finding alarming gaps on topics like harm minimization and responsible disclosure. To alleviate these concerns, we provide concrete recommendations to help NLP researchers navigate this space more ethically, bridging the gap between traditional cybersecurity and NLP ethics, which we frame as ``white hat NLP''. The goal of this work is to help cultivate an intentional culture of ethical research for those working in NLP Security.
Authors: Llewyn Salt, Marcus Gallagher
Abstract: Hyperparameter optimisation (HPO) is crucial for achieving strong performance in reinforcement learning (RL), as RL algorithms are inherently sensitive to hyperparameter settings. Probabilistic Curriculum Learning (PCL) is a curriculum learning strategy designed to improve RL performance by structuring the agent's learning process, yet effective hyperparameter tuning remains challenging and computationally demanding. In this paper, we provide an empirical analysis of hyperparameter interactions and their effects on the performance of a PCL algorithm within standard RL tasks, including point-maze navigation and DC motor control. Using the AlgOS framework integrated with Optuna's Tree-Structured Parzen Estimator (TPE), we present strategies to refine hyperparameter search spaces, enhancing optimisation efficiency. Additionally, we introduce a novel SHAP-based interpretability approach tailored specifically for analysing hyperparameter impacts, offering clear insights into how individual hyperparameters and their interactions influence RL performance. Our work contributes practical guidelines and interpretability tools that significantly improve the effectiveness and computational feasibility of hyperparameter optimisation in reinforcement learning.
Authors: Pedro Hermosilla, Christian Stippel, Leon Sick
Abstract: Self-supervised learning has transformed 2D computer vision by enabling models trained on large, unannotated datasets to provide versatile off-the-shelf features that perform similarly to models trained with labels. However, in 3D scene understanding, self-supervised methods are typically only used as a weight initialization step for task-specific fine-tuning, limiting their utility for general-purpose feature extraction. This paper addresses this shortcoming by proposing a robust evaluation protocol specifically designed to assess the quality of self-supervised features for 3D scene understanding. Our protocol uses multi-resolution feature sampling of hierarchical models to create rich point-level representations that capture the semantic capabilities of the model and, hence, are suitable for evaluation with linear probing and nearest-neighbor methods. Furthermore, we introduce the first self-supervised model that performs similarly to supervised models when only off-the-shelf features are used in a linear probing setup. In particular, our model is trained natively in 3D with a novel self-supervised approach based on a Masked Scene Modeling objective, which reconstructs deep features of masked patches in a bottom-up manner and is specifically tailored to hierarchical 3D models. Our experiments not only demonstrate that our method achieves competitive performance to supervised models, but also surpasses existing self-supervised approaches by a large margin. The model and training code can be found at our Github repository (https://github.com/phermosilla/msm).
Authors: Niccol\`o Turcato, Marco Cal\`i, Alberto Dalla Libera, Giulio Giacomuzzo, Ruggero Carli, Diego Romeres
Abstract: This short paper describes our proposed solution for the third edition of the "AI Olympics with RealAIGym" competition, held at ICRA 2025. We employed Monte-Carlo Probabilistic Inference for Learning Control (MC-PILCO), an MBRL algorithm recognized for its exceptional data efficiency across various low-dimensional robotic tasks, including cart-pole, ball \& plate, and Furuta pendulum systems. MC-PILCO optimizes a system dynamics model using interaction data, enabling policy refinement through simulation rather than direct system data optimization. This approach has proven highly effective in physical systems, offering greater data efficiency than Model-Free (MF) alternatives. Notably, MC-PILCO has previously won the first two editions of this competition, demonstrating its robustness in both simulated and real-world environments. Besides briefly reviewing the algorithm, we discuss the most critical aspects of the MC-PILCO implementation in the tasks at hand: learning a global policy for the pendubot and acrobot systems.
Authors: Wenfeng Feng, Guoying Sun
Abstract: In this paper, we propose EDIT (Encoder-Decoder Image Transformer), a novel architecture designed to mitigate the attention sink phenomenon observed in Vision Transformer models. Attention sink occurs when an excessive amount of attention is allocated to the [CLS] token, distorting the model's ability to effectively process image patches. To address this, we introduce a layer-aligned encoder-decoder architecture, where the encoder utilizes self-attention to process image patches, while the decoder uses cross-attention to focus on the [CLS] token. Unlike traditional encoder-decoder framework, where the decoder depends solely on high-level encoder representations, EDIT allows the decoder to extract information starting from low-level features, progressively refining the representation layer by layer. EDIT is naturally interpretable demonstrated through sequential attention maps, illustrating the refined, layer-by-layer focus on key image features. Experiments on ImageNet-1k and ImageNet-21k, along with transfer learning tasks, show that EDIT achieves consistent performance improvements over DeiT3 models. These results highlight the effectiveness of EDIT's design in addressing attention sink and improving visual feature extraction.
Authors: Yuankun Xie, Ruibo Fu, Zhiyong Wang, Xiaopeng Wang, Songjun Cao, Long Ma, Haonan Cheng, Long Ye
Abstract: The rapid advancement of audio generation technologies has escalated the risks of malicious deepfake audio across speech, sound, singing voice, and music, threatening multimedia security and trust. While existing countermeasures (CMs) perform well in single-type audio deepfake detection (ADD), their performance declines in cross-type scenarios. This paper is dedicated to studying the alltype ADD task. We are the first to comprehensively establish an all-type ADD benchmark to evaluate current CMs, incorporating cross-type deepfake detection across speech, sound, singing voice, and music. Then, we introduce the prompt tuning self-supervised learning (PT-SSL) training paradigm, which optimizes SSL frontend by learning specialized prompt tokens for ADD, requiring 458x fewer trainable parameters than fine-tuning (FT). Considering the auditory perception of different audio types,we propose the wavelet prompt tuning (WPT)-SSL method to capture type-invariant auditory deepfake information from the frequency domain without requiring additional training parameters, thereby enhancing performance over FT in the all-type ADD task. To achieve an universally CM, we utilize all types of deepfake audio for co-training. Experimental results demonstrate that WPT-XLSR-AASIST achieved the best performance, with an average EER of 3.58% across all evaluation sets. The code is available online.
Authors: Leon Reicherts, Zelun Tony Zhang, Elisabeth von Oswald, Yuanting Liu, Yvonne Rogers, Mariam Hassib
Abstract: How can we design AI tools that effectively support human decision-making by complementing and enhancing users' reasoning processes? Common recommendation-centric approaches face challenges such as inappropriate reliance or a lack of integration with users' decision-making processes. Here, we explore an alternative interaction model in which the AI outputs build upon users' own decision-making rationales. We compare this approach, which we call ExtendAI, with a recommendation-based AI. Participants in our mixed-methods user study interacted with both AIs as part of an investment decision-making task. We found that the AIs had different impacts, with ExtendAI integrating better into the decision-making process and people's own thinking and leading to slightly better outcomes. RecommendAI was able to provide more novel insights while requiring less cognitive effort. We discuss the implications of these and other findings along with three tensions of AI-assisted decision-making which our study revealed.
Authors: Shuoshuo Xu, Kai Zhao, James Loney, Zili Li, Andrea Visentin
Abstract: Effective and rapid evaluation of pavement surface condition is critical for prioritizing maintenance, ensuring transportation safety, and minimizing vehicle wear and tear. While conventional manual inspections suffer from subjectivity, existing machine learning-based methods are constrained by their reliance on large and high-quality labeled datasets, which require significant resources and limit adaptability across varied road conditions. The revolutionary advancements in Large Language Models (LLMs) present significant potential for overcoming these challenges. In this study, we propose an innovative automated zero-shot learning approach that leverages the image recognition and natural language understanding capabilities of LLMs to assess road conditions effectively. Multiple LLM-based assessment models were developed, employing prompt engineering strategies aligned with the Pavement Surface Condition Index (PSCI) standards. These models' accuracy and reliability were evaluated against official PSCI results, with an optimized model ultimately selected. Extensive tests benchmarked the optimized model against evaluations from various levels experts using Google Street View road images. The results reveal that the LLM-based approach can effectively assess road conditions, with the optimized model -employing comprehensive and structured prompt engineering strategies -outperforming simpler configurations by achieving high accuracy and consistency, even surpassing expert evaluations. Moreover, successfully applying the optimized model to Google Street View images demonstrates its potential for future city-scale deployments. These findings highlight the transformative potential of LLMs in automating road damage evaluations and underscore the pivotal role of detailed prompt engineering in achieving reliable assessments.
Authors: Willian Soares Gir\~ao, Nicoletta Risi, Elisabetta Chicca
Abstract: Understanding how biological neural networks are shaped via local plasticity mechanisms can lead to energy-efficient and self-adaptive information processing systems, which promises to mitigate some of the current roadblocks in edge computing systems. While biology makes use of spikes to seamless use both spike timing and mean firing rate to modulate synaptic strength, most models focus on one of the two. In this work, we present a Hebbian local learning rule that models synaptic modification as a function of calcium traces tracking neuronal activity. We show how the rule reproduces results from spike time and spike rate protocols from neuroscientific studies. Moreover, we use the model to train spiking neural networks on MNIST digit recognition to show and explain what sort of mechanisms are needed to learn real-world patterns. We show how our model is sensitive to correlated spiking activity and how this enables it to modulate the learning rate of the network without altering the mean firing rate of the neurons nor the hyparameters of the learning rule. To the best of our knowledge, this is the first work that showcases how spike timing and rate can be complementary in their role of shaping the connectivity of spiking neural networks.
Authors: Ali Goli, Mahdieh Alizadeh, Hadi Sadoghi Yazdi
Abstract: Manifold learning techniques, such as Locally linear embedding (LLE), are designed to preserve the local neighborhood structures of high-dimensional data during dimensionality reduction. Traditional LLE employs Euclidean distance to define neighborhoods, which can struggle to capture the intrinsic geometric relationships within complex data. A novel approach, Adaptive locally linear embedding(ALLE), is introduced to address this limitation by incorporating a dynamic, data-driven metric that enhances topological preservation. This method redefines the concept of proximity by focusing on topological neighborhood inclusion rather than fixed distances. By adapting the metric based on the local structure of the data, it achieves superior neighborhood preservation, particularly for datasets with complex geometries and high-dimensional structures. Experimental results demonstrate that ALLE significantly improves the alignment between neighborhoods in the input and feature spaces, resulting in more accurate and topologically faithful embeddings. This approach advances manifold learning by tailoring distance metrics to the underlying data, providing a robust solution for capturing intricate relationships in high-dimensional datasets.
Authors: Angela Lopez-Cardona, Sebastian Idesis, Ioannis Arapakis
Abstract: Recently, the integration of cognitive neuroscience in Natural Language Processing (NLP) has gained significant attention. This article provides a critical and timely overview of recent advancements in leveraging cognitive signals, particularly Eye-tracking (ET) signals, to enhance Language Models (LMs) and Multimodal Large Language Models (MLLMs). By incorporating user-centric cognitive signals, these approaches address key challenges, including data scarcity and the environmental costs of training large-scale models. Cognitive signals enable efficient data augmentation, faster convergence, and improved human alignment. The review emphasises the potential of ET data in tasks like Visual Question Answering (VQA) and mitigating hallucinations in MLLMs, and concludes by discussing emerging challenges and research trends.
Authors: Diljeet Jagpal, Xi Chen, Vinay P. Namboodiri
Abstract: Zero-shot, training-free, image-based text-to-video generation is an emerging area that aims to generate videos using existing image-based diffusion models. Current methods in this space require specific architectural changes to image generation models, which limit their adaptability and scalability. In contrast to such methods, we provide a model-agnostic approach. We use intersections in diffusion trajectories, working only with the latent values. We could not obtain localized frame-wise coherence and diversity using only the intersection of trajectories. Thus, we instead use a grid-based approach. An in-context trained LLM is used to generate coherent frame-wise prompts; another is used to identify differences between frames. Based on these, we obtain a CLIP-based attention mask that controls the timing of switching the prompts for each grid cell. Earlier switching results in higher variance, while later switching results in more coherence. Therefore, our approach can ensure appropriate control between coherence and variance for the frames. Our approach results in state-of-the-art performance while being more flexible when working with diverse image-generation models. The empirical analysis using quantitative metrics and user studies confirms our model's superior temporal consistency, visual fidelity and user satisfaction, thus providing a novel way to obtain training-free, image-based text-to-video generation.
Authors: Seunghyeok Back, Joosoon Lee, Kangmin Kim, Heeseon Rho, Geonhyup Lee, Raeyoung Kang, Sangbeom Lee, Sangjun Noh, Youngjin Lee, Taeyeop Lee, Kyoobin Lee
Abstract: Robust grasping in cluttered environments remains an open challenge in robotics. While benchmark datasets have significantly advanced deep learning methods, they mainly focus on simplistic scenes with light occlusion and insufficient diversity, limiting their applicability to practical scenarios. We present GraspClutter6D, a large-scale real-world grasping dataset featuring: (1) 1,000 highly cluttered scenes with dense arrangements (14.1 objects/scene, 62.6\% occlusion), (2) comprehensive coverage across 200 objects in 75 environment configurations (bins, shelves, and tables) captured using four RGB-D cameras from multiple viewpoints, and (3) rich annotations including 736K 6D object poses and 9.3B feasible robotic grasps for 52K RGB-D images. We benchmark state-of-the-art segmentation, object pose estimation, and grasping detection methods to provide key insights into challenges in cluttered environments. Additionally, we validate the dataset's effectiveness as a training resource, demonstrating that grasping networks trained on GraspClutter6D significantly outperform those trained on existing datasets in both simulation and real-world experiments. The dataset, toolkit, and annotation tools are publicly available on our project website: https://sites.google.com/view/graspclutter6d.
Authors: Seungwon Lim, Seungbeen Lee, Dongjun Min, Youngjae Yu
Abstract: Artificial agents are increasingly central to complex interactions and decision-making tasks, yet aligning their behaviors with desired human values remains an open challenge. In this work, we investigate how human-like personality traits influence agent behavior and performance within text-based interactive environments. We introduce PANDA: PersonalityAdapted Neural Decision Agents, a novel method for projecting human personality traits onto agents to guide their behavior. To induce personality in a text-based game agent, (i) we train a personality classifier to identify what personality type the agent's actions exhibit, and (ii) we integrate the personality profiles directly into the agent's policy-learning pipeline. By deploying agents embodying 16 distinct personality types across 25 text-based games and analyzing their trajectories, we demonstrate that an agent's action decisions can be guided toward specific personality profiles. Moreover, certain personality types, such as those characterized by higher levels of Openness, display marked advantages in performance. These findings underscore the promise of personality-adapted agents for fostering more aligned, effective, and human-centric decision-making in interactive environments.
Authors: Mingbo Li, Liying Liu, Ye Luo
Abstract: Convolutional neural networks have become increasingly deep and complex, leading to higher computational costs. While tropical convolutional neural networks (TCNNs) reduce multiplications, they underperform compared to standard CNNs. To address this, we propose two new variants - compound TCNN (cTCNN) and parallel TCNN (pTCNN)-that use combinations of tropical min-plus and max-plus kernels to replace traditional convolution kernels. This reduces multiplications and balances efficiency with performance. Experiments on various datasets show that cTCNN and pTCNN match or exceed the performance of other CNN methods. Combining these with conventional CNNs in deeper architectures also improves performance. We are further exploring simplified TCNN architectures that reduce parameters and multiplications with minimal accuracy loss, aiming for efficient and effective models.
Authors: Wuyang Liu, Yi Chai, Yongpeng Yan, Yanzhen Ren
Abstract: Audio-visual event localization (AVEL) plays a critical role in multimodal scene understanding. While existing datasets for AVEL predominantly comprise landscape-oriented long videos with clean and simple audio context, short videos have become the primary format of online video content due to the the proliferation of smartphones. Short videos are characterized by portrait-oriented framing and layered audio compositions (e.g., overlapping sound effects, voiceovers, and music), which brings unique challenges unaddressed by conventional methods. To this end, we introduce AVE-PM, the first AVEL dataset specifically designed for portrait mode short videos, comprising 25,335 clips that span 86 fine-grained categories with frame-level annotations. Beyond dataset creation, our empirical analysis shows that state-of-the-art AVEL methods suffer an average 18.66% performance drop during cross-mode evaluation. Further analysis reveals two key challenges of different video formats: 1) spatial bias from portrait-oriented framing introduces distinct domain priors, and 2) noisy audio composition compromise the reliability of audio modality. To address these issues, we investigate optimal preprocessing recipes and the impact of background music for AVEL on portrait mode videos. Experiments show that these methods can still benefit from tailored preprocessing and specialized model design, thus achieving improved performance. This work provides both a foundational benchmark and actionable insights for advancing AVEL research in the era of mobile-centric video content. Dataset and code will be released.
Authors: Jiawei Mao, Yuhan Wang, Yucheng Tang, Daguang Xu, Kang Wang, Yang Yang, Zongwei Zhou, Yuyin Zhou
Abstract: This paper presents MedSegFactory, a versatile medical synthesis framework that generates high-quality paired medical images and segmentation masks across modalities and tasks. It aims to serve as an unlimited data repository, supplying image-mask pairs to enhance existing segmentation tools. The core of MedSegFactory is a dual-stream diffusion model, where one stream synthesizes medical images and the other generates corresponding segmentation masks. To ensure precise alignment between image-mask pairs, we introduce Joint Cross-Attention (JCA), enabling a collaborative denoising paradigm by dynamic cross-conditioning between streams. This bidirectional interaction allows both representations to guide each other's generation, enhancing consistency between generated pairs. MedSegFactory unlocks on-demand generation of paired medical images and segmentation masks through user-defined prompts that specify the target labels, imaging modalities, anatomical regions, and pathological conditions, facilitating scalable and high-quality data generation. This new paradigm of medical image synthesis enables seamless integration into diverse medical imaging workflows, enhancing both efficiency and accuracy. Extensive experiments show that MedSegFactory generates data of superior quality and usability, achieving competitive or state-of-the-art performance in 2D and 3D segmentation tasks while addressing data scarcity and regulatory constraints.
Authors: Miro Miranda, Francisco Mena, Andreas Dengel
Abstract: Missing instances in time series data impose a significant challenge to deep learning models, particularly in regression tasks. In the Earth Observation field, satellite failure or cloud occlusion frequently results in missing time-steps, introducing uncertainties in the predicted output and causing a decline in predictive performance. While many studies address missing time-steps through data augmentation to improve model robustness, the uncertainty arising at the input level is commonly overlooked. To address this gap, we introduce Monte Carlo Temporal Dropout (MC-TD), a method that explicitly accounts for input-level uncertainty by randomly dropping time-steps during inference using a predefined dropout ratio, thereby simulating the effect of missing data. To bypass the need for costly searches for the optimal dropout ratio, we extend this approach with Monte Carlo Concrete Temporal Dropout (MC-ConcTD), a method that learns the optimal dropout distribution directly. Both MC-TD and MC-ConcTD are applied during inference, leveraging Monte Carlo sampling for uncertainty quantification. Experiments on three EO time-series datasets demonstrate that MC-ConcTD improves predictive performance and uncertainty calibration compared to existing approaches. Additionally, we highlight the advantages of adaptive dropout tuning over manual selection, making uncertainty quantification more robust and accessible for EO applications.
Authors: Anisa V. Prasad, Tejas Sudharshan Mathai, Pritam Mukherjee, Jianfei Liu, Ronald M. Summers
Abstract: An accurate segmentation of the pancreas on CT is crucial to identify pancreatic pathologies and extract imaging-based biomarkers. However, prior research on pancreas segmentation has primarily focused on modifying the segmentation model architecture or utilizing pre- and post-processing techniques. In this article, we investigate the utility of anatomical priors to enhance the segmentation performance of the pancreas. Two 3D full-resolution nnU-Net models were trained, one with 8 refined labels from the public PANORAMA dataset, and another that combined them with labels derived from the public TotalSegmentator (TS) tool. The addition of anatomical priors resulted in a 6\% increase in Dice score ($p < .001$) and a 36.5 mm decrease in Hausdorff distance for pancreas segmentation ($p < .001$). Moreover, the pancreas was always detected when anatomy priors were used, whereas there were 8 instances of failed detections without their use. The use of anatomy priors shows promise for pancreas segmentation and subsequent derivation of imaging biomarkers.
Authors: Tejas Sudharshan Mathai, Benjamin Hou, Ronald M. Summers
Abstract: In the U.S., lung cancer is the second major cause of death. Early detection of suspicious lung nodules is crucial for patient treatment planning, management, and improving outcomes. Many approaches for lung nodule segmentation and volumetric analysis have been proposed, but few have looked at longitudinal changes in total lung tumor burden. In this work, we trained two 3D models (nnUNet) with and without anatomical priors to automatically segment lung lesions and quantified total lesion burden for each patient. The 3D model without priors significantly outperformed ($p < .001$) the model trained with anatomy priors. For detecting clinically significant lesions $>$ 1cm, a precision of 71.3\%, sensitivity of 68.4\%, and F1-score of 69.8\% was achieved. For segmentation, a Dice score of 77.1 $\pm$ 20.3 and Hausdorff distance error of 11.7 $\pm$ 24.1 mm was obtained. The median lesion burden was 6.4 cc (IQR: 2.1, 18.1) and the median volume difference between manual and automated measurements was 0.02 cc (IQR: -2.8, 1.2). Agreements were also evaluated with linear regression and Bland-Altman plots. The proposed approach can produce a personalized evaluation of the total tumor burden for a patient and facilitate interval change tracking over time.
Authors: Sergio Romero-Tapiador, Ruben Tolosana, Blanca Lacruz-Pleguezuelos, Laura Judith Marcos Zambrano, Guadalupe X. Baz\'an, Isabel Espinosa-Salinas, Julian Fierrez, Javier Ortega-Garcia, Enrique Carrillo de Santa Pau, Aythami Morales
Abstract: Automatic dietary assessment based on food images remains a challenge, requiring precise food detection, segmentation, and classification. Vision-Language Models (VLMs) offer new possibilities by integrating visual and textual reasoning. In this study, we evaluate six state-of-the-art VLMs (ChatGPT, Gemini, Claude, Moondream, DeepSeek, and LLaVA), analyzing their capabilities in food recognition at different levels. For the experimental framework, we introduce the FoodNExTDB, a unique food image database that contains 9,263 expert-labeled images across 10 categories (e.g., "protein source"), 62 subcategories (e.g., "poultry"), and 9 cooking styles (e.g., "grilled"). In total, FoodNExTDB includes 50k nutritional labels generated by seven experts who manually annotated all images in the database. Also, we propose a novel evaluation metric, Expert-Weighted Recall (EWR), that accounts for the inter-annotator variability. Results show that closed-source models outperform open-source ones, achieving over 90% EWR in recognizing food products in images containing a single product. Despite their potential, current VLMs face challenges in fine-grained food recognition, particularly in distinguishing subtle differences in cooking styles and visually similar food items, which limits their reliability for automatic dietary assessment. The FoodNExTDB database is publicly available at https://github.com/AI4Food/FoodNExtDB.
Authors: Bodong Chen
Abstract: As generative AI rapidly integrates into educational infrastructures worldwide, it transforms how knowledge gets created, validated, and shared, yet current discourse inadequately addresses its implications as epistemic infrastructure mediating teaching and learning. This paper investigates how AI systems function as epistemic infrastructures in education and their impact on human epistemic agency. Adopting a situated cognition perspective and following a value-sensitive design approach, the study conducts a technical investigation of two representative AI systems in educational settings, analyzing their impact on teacher practice across three dimensions: affordances for skilled epistemic actions, support for epistemic sensitivity, and implications for long-term habit formation. The analysis reveals that current AI systems inadequately support teachers' skilled epistemic actions, insufficiently foster epistemic sensitivity, and potentially cultivate problematic habits that prioritize efficiency over epistemic agency. To address these challenges, the paper recommends recognizing the infrastructural transformation occurring in education, developing AI environments that stimulate skilled actions while upholding epistemic norms, and involving educators in AI design processes -- recommendations aimed at fostering AI integration that aligns with core educational values and maintains human epistemic agency.
Authors: Zhixuan Lin, Johan Obando-Ceron, Xu Owen He, Aaron Courville
Abstract: The recently proposed Forgetting Transformer (FoX) incorporates a forget gate into softmax attention and has shown consistently better or on-par performance compared to the standard RoPE-based Transformer. Notably, many attention heads in FoX tend to forget quickly, causing their output at each timestep to rely primarily on the local context. Based on this observation, we propose Adaptive Computation Pruning (ACP) for FoX, a method that dynamically prunes computations involving input-output dependencies that are strongly decayed by the forget gate. This is achieved using a dynamically set pruning threshold that ensures that the pruned attention weights remain negligible. We apply ACP to language model pretraining with FoX and show it consistently reduces the number of FLOPs in softmax attention by around 70% across different model sizes and context lengths, resulting in a roughly 10% to 35% improvement in training throughput. Furthermore, longer context lengths yield greater computational savings. All these speed improvements are achieved without any performance degradation. We also perform several analyses to provide deeper insights into our method, such as examining the pruning patterns and analyzing the distribution of FLOP savings across different attention heads. Our code is available at https://github.com/zhixuan-lin/arctic-fox.
Authors: Thomas Kerdreux, Alexandre Tuel, Quentin Febvre, Alexis Mouche, Bertrand Chapron
Abstract: Self-supervised learning (SSL) has enabled the development of vision foundation models for Earth Observation (EO), demonstrating strong transferability across diverse remote sensing tasks. While prior work has focused on network architectures and training strategies, the role of dataset curation, especially in balancing and diversifying pre-training datasets, remains underexplored. In EO, this challenge is amplified by the redundancy and heavy-tailed distributions common in satellite imagery, which can lead to biased representations and inefficient training. In this work, we propose a dynamic dataset pruning strategy designed to improve SSL pre-training by maximizing dataset diversity and balance. Our method iteratively refines the training set without requiring a pre-existing feature extractor, making it well-suited for domains where curated datasets are limited or unavailable. We demonstrate our approach on the Sentinel-1 Wave Mode (WV) Synthetic Aperture Radar (SAR) archive, a challenging dataset dominated by ocean observations. We train models from scratch on the entire Sentinel-1 WV archive spanning 10 years. Across three downstream tasks, our results show that dynamic pruning improves both computational efficiency and representation quality, leading to stronger transferability. We also release the weights of Nereus-SAR-1, the first model in the Nereus family, a series of foundation models for ocean observation and analysis using SAR imagery, at github.com/galeio-research/nereus-sar-models/.
Authors: Vladimir Bataev
Abstract: Training speech recognition systems on noisy transcripts is a significant challenge in industrial pipelines, where datasets are enormous and ensuring accurate transcription for every instance is difficult. In this work, we introduce novel loss functions to mitigate the impact of transcription errors in RNN-Transducer models. Our Star-Transducer loss addresses deletion errors by incorporating "skip frame" transitions in the loss lattice, restoring over 90% of the system's performance compared to models trained with accurate transcripts. The Bypass-Transducer loss uses "skip token" transitions to tackle insertion errors, recovering more than 60% of the quality. Finally, the Target-Robust Transducer loss merges these approaches, offering robust performance against arbitrary errors. Experimental results demonstrate that the Target-Robust Transducer loss significantly improves RNN-T performance on noisy data by restoring over 70% of the quality compared to well-transcribed data.
Authors: Sanyam Paresh Shah, Abdullah Mamun, Shovito Barua Soumma, Hassan Ghasemzadeh
Abstract: Metabolic Syndrome (MetS) is a cluster of interrelated risk factors that significantly increases the risk of cardiovascular diseases and type 2 diabetes. Despite its global prevalence, accurate prediction of MetS remains challenging due to issues such as class imbalance, data scarcity, and methodological inconsistencies in existing studies. In this paper, we address these challenges by systematically evaluating and optimizing machine learning (ML) models for MetS prediction, leveraging advanced data balancing techniques and counterfactual analysis. Multiple ML models, including XGBoost, Random Forest, TabNet, etc., were trained and compared under various data balancing techniques such as random oversampling (ROS), SMOTE, ADASYN, and CTGAN. Additionally, we introduce MetaBoost, a novel hybrid framework that integrates SMOTE, ADASYN, and CTGAN, optimizing synthetic data generation through weighted averaging and iterative weight tuning to enhance the model's performance (achieving a 1.14% accuracy improvement over individual balancing techniques). A comprehensive counterfactual analysis is conducted to quantify feature-level changes required to shift individuals from high-risk to low-risk categories. The results indicate that blood glucose (50.3%) and triglycerides (46.7%) were the most frequently modified features, highlighting their clinical significance in MetS risk reduction. Additionally, probabilistic analysis shows elevated blood glucose (85.5% likelihood) and triglycerides (74.9% posterior probability) as the strongest predictors. This study not only advances the methodological rigor of MetS prediction but also provides actionable insights for clinicians and researchers, highlighting the potential of ML in mitigating the public health burden of metabolic syndrome.
Authors: Omar Alama, Avigyan Bhattacharya, Haoyang He, Seungchan Kim, Yuheng Qiu, Wenshan Wang, Cherie Ho, Nikhil Keetha, Sebastian Scherer
Abstract: Open-set semantic mapping is crucial for open-world robots. Current mapping approaches either are limited by the depth range or only map beyond-range entities in constrained settings, where overall they fail to combine within-range and beyond-range observations. Furthermore, these methods make a trade-off between fine-grained semantics and efficiency. We introduce RayFronts, a unified representation that enables both dense and beyond-range efficient semantic mapping. RayFronts encodes task-agnostic open-set semantics to both in-range voxels and beyond-range rays encoded at map boundaries, empowering the robot to reduce search volumes significantly and make informed decisions both within & beyond sensory range, while running at 8.84 Hz on an Orin AGX. Benchmarking the within-range semantics shows that RayFronts's fine-grained image encoding provides 1.34x zero-shot 3D semantic segmentation performance while improving throughput by 16.5x. Traditionally, online mapping performance is entangled with other system components, complicating evaluation. We propose a planner-agnostic evaluation framework that captures the utility for online beyond-range search and exploration, and show RayFronts reduces search volume 2.2x more efficiently than the closest online baselines.
Authors: Bibek Paudel, Alexander Lyzhov, Preetam Joshi, Puneet Anand
Abstract: This paper introduces a comprehensive system for detecting hallucinations in large language model (LLM) outputs in enterprise settings. We present a novel taxonomy of LLM responses specific to hallucination in enterprise applications, categorizing them into context-based, common knowledge, enterprise-specific, and innocuous statements. Our hallucination detection model HDM-2 validates LLM responses with respect to both context and generally known facts (common knowledge). It provides both hallucination scores and word-level annotations, enabling precise identification of problematic content. To evaluate it on context-based and common-knowledge hallucinations, we introduce a new dataset HDMBench. Experimental results demonstrate that HDM-2 out-performs existing approaches across RagTruth, TruthfulQA, and HDMBench datasets. This work addresses the specific challenges of enterprise deployment, including computational efficiency, domain specialization, and fine-grained error identification. Our evaluation dataset, model weights, and inference code are publicly available.
Authors: Atharva Pandey, Kshitij Dubey, Rahul Sharma, Amit Sharma
Abstract: Despite great performance on Olympiad-level reasoning problems, frontier large language models can still struggle on high school math when presented with novel problems outside standard benchmarks. Going beyond final accuracy, we propose a deductive consistency metric to analyze chain-of-thought output from language models (LMs).Formally, deductive reasoning involves two subtasks: understanding a set of input premises and inferring the conclusions that follow from them. The proposed metric studies LMs' performance on these subtasks, with the goal of explaining LMs' reasoning errors on novel problems: how well do LMs understand input premises with increasing context lengths, and how well can they infer conclusions over multiple reasoning hops? Since existing benchmarks may be memorized, we develop a pipeline to evaluate LMs' deductive consistency on novel, perturbed versions of benchmark problems. On novel grade school math problems (GSM-8k), we find that LMs are fairly robust to increasing number of input premises, but suffer significant accuracy decay as the number of reasoning hops is increased. Interestingly, these errors are masked in the original benchmark as all models achieve near 100% accuracy. As we increase the number of solution steps using a synthetic dataset, prediction over multiple hops still remains the major source of error compared to understanding input premises. Other factors, such as shifts in language style or natural propagation of early errors do not explain the trends. Our analysis provides a new view to characterize LM reasoning -- as computations over a window of input premises and reasoning hops -- that can provide unified evaluation across problem domains.
Authors: Gabriel Grand, Joshua B. Tenenbaum, Vikash K. Mansinghka, Alexander K. Lew, Jacob Andreas
Abstract: While test-time reasoning enables language models to tackle complex tasks, searching or planning in natural language can be slow, costly, and error-prone. But even when LMs struggle to emulate the precise reasoning steps needed to solve a problem, they often excel at describing its abstract structure--both how to verify solutions and how to search for them. This paper introduces DisCIPL, a method for "self-steering" LMs where a Planner model generates a task-specific inference program that is executed by a population of Follower models. Our approach equips LMs with the ability to write recursive search procedures that guide LM inference, enabling new forms of verifiable and efficient reasoning. When instantiated with a small Follower (e.g., Llama-3.2-1B), DisCIPL matches (and sometimes outperforms) much larger models, including GPT-4o and o1, on challenging constrained generation tasks. In decoupling planning from execution, our work opens up a design space of highly-parallelized Monte Carlo inference strategies that outperform standard best-of-N sampling, require no finetuning, and can be implemented automatically by existing LMs.
Authors: Elan Markowitz, Krupa Galiya, Greg Ver Steeg, Aram Galstyan
Abstract: Knowledge graphs have emerged as a popular method for injecting up-to-date, factual knowledge into large language models (LLMs). This is typically achieved by converting the knowledge graph into text that the LLM can process in context. While multiple methods of encoding knowledge graphs have been proposed, the impact of this textualization process on LLM performance remains under-explored. We introduce KG-LLM-Bench, a comprehensive and extensible benchmark spanning five knowledge graph understanding tasks, and evaluate how different encoding strategies affect performance across various base models. Our extensive experiments with seven language models and five textualization strategies provide insights for optimizing LLM performance on KG reasoning tasks.
Authors: Alexander Rubinstein, Ameya Prabhu, Matthias Bethge, Seong Joon Oh
Abstract: Object-centric learning (OCL) seeks to learn representations that only encode an object, isolated from other objects or background cues in a scene. This approach underpins various aims, including out-of-distribution (OOD) generalization, sample-efficient composition, and modeling of structured environments. Most research has focused on developing unsupervised mechanisms that separate objects into discrete slots in the representation space, evaluated using unsupervised object discovery. However, with recent sample-efficient segmentation models, we can separate objects in the pixel space and encode them independently. This achieves remarkable zero-shot performance on OOD object discovery benchmarks, is scalable to foundation models, and can handle a variable number of slots out-of-the-box. Hence, the goal of OCL methods to obtain object-centric representations has been largely achieved. Despite this progress, a key question remains: How does the ability to separate objects within a scene contribute to broader OCL objectives, such as OOD generalization? We address this by investigating the OOD generalization challenge caused by spurious background cues through the lens of OCL. We propose a novel, training-free probe called $\textbf{Object-Centric Classification with Applied Masks (OCCAM)}$, demonstrating that segmentation-based encoding of individual objects significantly outperforms slot-based OCL methods. However, challenges in real-world applications remain. We provide the toolbox for the OCL community to use scalable object-centric representations, and focus on practical applications and fundamental questions, such as understanding object perception in human cognition. Our code is available $\href{https://github.com/AlexanderRubinstein/OCCAM}{here}$.
Authors: Nikhil Shivakumar Nayak, Krishnateja Killamsetty, Ligong Han, Abhishek Bhandwaldar, Prateek Chanda, Kai Xu, Hao Wang, Aldo Pareja, Oleg Silkin, Mustafa Eyceoz, Akash Srivastava
Abstract: Continual learning in large language models (LLMs) is prone to catastrophic forgetting, where adapting to new tasks significantly degrades performance on previously learned ones. Existing methods typically rely on low-rank, parameter-efficient updates that limit the model's expressivity and introduce additional parameters per task, leading to scalability issues. To address these limitations, we propose a novel continual full fine-tuning approach leveraging adaptive singular value decomposition (SVD). Our method dynamically identifies task-specific low-rank parameter subspaces and constrains updates to be orthogonal to critical directions associated with prior tasks, thus effectively minimizing interference without additional parameter overhead or storing previous task gradients. We evaluate our approach extensively on standard continual learning benchmarks using both encoder-decoder (T5-Large) and decoder-only (LLaMA-2 7B) models, spanning diverse tasks including classification, generation, and reasoning. Empirically, our method achieves state-of-the-art results, up to 7% higher average accuracy than recent baselines like O-LoRA, and notably maintains the model's general linguistic capabilities, instruction-following accuracy, and safety throughout the continual learning process by reducing forgetting to near-negligible levels. Our adaptive SVD framework effectively balances model plasticity and knowledge retention, providing a practical, theoretically grounded, and computationally scalable solution for continual learning scenarios in large language models.
Authors: Srijoni Majumdar, Edith Elkind, Evangelos Pournaras
Abstract: Scaling up deliberative and voting participation is a longstanding endeavor -- a cornerstone for direct democracy and legitimate collective choice. Recent breakthroughs in generative artificial intelligence (AI) and large language models (LLMs) unravel new capabilities for AI personal assistants to overcome cognitive bandwidth limitations of humans, providing decision support or even direct representation of human voters at large scale. However, the quality of this representation and what underlying biases manifest when delegating collective decision-making to LLMs is an alarming and timely challenge to tackle. By rigorously emulating with high realism more than >50K LLM voting personas in 306 real-world voting elections, we disentangle the nature of different biases in LLMS (GPT 3, GPT 3.5, and Llama2). Complex preferential ballot formats exhibit significant inconsistencies compared to simpler majoritarian elections that show higher consistency. Strikingly though, by demonstrating for the first time in real-world a proportional representation of voters in direct democracy, we are also able to show that fair ballot aggregation methods, such as equal shares, prove to be a win-win: fairer voting outcomes for humans with fairer AI representation, especially for voters who are likely to abstain. This novel underlying relationship proves paramount for democratic resilience in progressives scenarios with low voters turnout and voter fatigue supported by AI representatives: abstained voters are mitigated by recovering highly representative voting outcomes that are fairer. These interdisciplinary insights provide remarkable foundations for science, policymakers, and citizens to develop safeguards and resilience for AI risks in democratic innovations.
Authors: Elias Hossain, Tasfia Nuzhat, Shamsul Masum, Shahram Rahimi, Noorbakhsh Amiri Golilarz
Abstract: Accurate classification of cancer-related medical abstracts is crucial for healthcare management and research. However, obtaining large, labeled datasets in the medical domain is challenging due to privacy concerns and the complexity of clinical data. This scarcity of annotated data impedes the development of effective machine learning models for cancer document classification. To address this challenge, we present a curated dataset of 1,874 biomedical abstracts, categorized into thyroid cancer, colon cancer, lung cancer, and generic topics. Our research focuses on leveraging this dataset to improve classification performance, particularly in data-scarce scenarios. We introduce a Residual Graph Attention Network (R-GAT) with multiple graph attention layers that capture the semantic information and structural relationships within cancer-related documents. Our R-GAT model is compared with various techniques, including transformer-based models such as Bidirectional Encoder Representations from Transformers (BERT), RoBERTa, and domain-specific models like BioBERT and Bio+ClinicalBERT. We also evaluated deep learning models (CNNs, LSTMs) and traditional machine learning models (Logistic Regression, SVM). Additionally, we explore ensemble approaches that combine deep learning models to enhance classification. Various feature extraction methods are assessed, including Term Frequency-Inverse Document Frequency (TF-IDF) with unigrams and bigrams, Word2Vec, and tokenizers from BERT and RoBERTa. The R-GAT model outperforms other techniques, achieving precision, recall, and F1 scores of 0.99, 0.97, and 0.98 for thyroid cancer; 0.96, 0.94, and 0.95 for colon cancer; 0.96, 0.99, and 0.97 for lung cancer; and 0.95, 0.96, and 0.95 for generic topics.
Authors: Zifeng Wang, Benjamin Danek, Ziwei Yang, Zheng Chen, Jimeng Sun
Abstract: Data science plays a critical role in biomedical research, but it requires professionals with expertise in coding and medical data analysis. Large language models (LLMs) have shown great potential in supporting medical tasks and performing well in general coding tests. However, existing evaluations fail to assess their capability in biomedical data science, particularly in handling diverse data types such as genomics and clinical datasets. To address this gap, we developed a benchmark of data science coding tasks derived from the analyses of 39 published studies. This benchmark comprises 293 coding tasks (128 in Python and 165 in R) performed on real-world TCGA-type genomics and clinical data. Our findings reveal that the vanilla prompting of LLMs yields suboptimal performances due to drawbacks in following input instructions, understanding target data, and adhering to standard analysis practices. Next, we benchmarked six cutting-edge LLMs and advanced adaptation methods, finding two methods to be particularly effective: chain-of-thought prompting, which provides a step-by-step plan for data analysis, which led to a 21% code accuracy improvement (56.6% versus 35.3%); and self-reflection, enabling LLMs to refine the buggy code iteratively, yielding an 11% code accuracy improvement (45.5% versus 34.3%). Building on these insights, we developed a platform that integrates LLMs into the data science workflow for medical professionals. In a user study with five medical professionals, we found that while LLMs cannot fully automate programming tasks, they significantly streamline the programming process. We found that 80% of their submitted code solutions were incorporated from LLM-generated code, with up to 96% reuse in some cases. Our analysis highlights the potential of LLMs to enhance data science efficiency in biomedical research when integrated into expert workflows.
Authors: Jiakang Yuan, Xiangchao Yan, Shiyang Feng, Bo Zhang, Tao Chen, Botian Shi, Wanli Ouyang, Yu Qiao, Lei Bai, Bowen Zhou
Abstract: The scientific research paradigm is undergoing a profound transformation owing to the development of Artificial Intelligence (AI). Recent works demonstrate that various AI-assisted research methods can largely improve research efficiency by improving data analysis, accelerating computation, and fostering novel idea generation. To further move towards the ultimate goal (i.e., automatic scientific research), in this paper, we introduce Dolphin, a closed-loop LLM-driven framework to enhance the automation level of scientific research. Dolphin first generates novel ideas based on feedback from previous experiments and relevant papers ranked by the topic and task attributes. Then, the generated ideas can be implemented using a code template refined and debugged with the designed exception-traceback-guided local code structure. Finally, Dolphin automatically analyzes the results of each idea and feeds the results back to the next round of idea generation. Experiments are conducted on the benchmark datasets of different topics and a subset of MLE-bench. Results show that Dolphin can continuously improve the performance of the input topic in a loop. We highlight that Dolphin can automatically propose methods that are comparable to the state-of-the-art in some tasks such as 3D point classification.
Authors: Marco Giunti
Abstract: This paper critically examines the recent publication "ChatGPT-4 in the Turing Test" by Restrepo Echavarr\'ia (2025), challenging its central claims regarding the absence of minimally serious test implementations and the conclusion that ChatGPT-4 fails the Turing Test. The analysis reveals that the criticisms based on rigid criteria and limited experimental data are not fully justified. More importantly, the paper makes several constructive contributions that enrich our understanding of Turing Test implementations. It demonstrates that two distinct formats--the three-player and two-player tests--are both valid, each with unique methodological implications. The work distinguishes between absolute criteria (reflecting an optimal 50% identification rate in a three-player format) and relative criteria (which measure how closely a machine's performance approximates that of a human), offering a more nuanced evaluation framework. Furthermore, the paper clarifies the probabilistic underpinnings of both test types by modeling them as Bernoulli experiments--correlated in the three-player version and uncorrelated in the two-player version. This formalization allows for a rigorous separation between the theoretical criteria for passing the test, defined in probabilistic terms, and the experimental data that require robust statistical methods for proper interpretation. In doing so, the paper not only refutes key aspects of the criticized study but also lays a solid foundation for future research on objective measures of how closely an AI's behavior aligns with, or deviates from, that of a human being.
Authors: Qiguang Chen, Libo Qin, Jinhao Liu, Dengyun Peng, Jiannan Guan, Peng Wang, Mengkang Hu, Yuhang Zhou, Te Gao, Wanxiang Che
Abstract: Recent advancements in reasoning with large language models (RLLMs), such as OpenAI-O1 and DeepSeek-R1, have demonstrated their impressive capabilities in complex domains like mathematics and coding. A central factor in their success lies in the application of long chain-of-thought (Long CoT) characteristics, which enhance reasoning abilities and enable the solution of intricate problems. However, despite these developments, a comprehensive survey on Long CoT is still lacking, limiting our understanding of its distinctions from traditional short chain-of-thought (Short CoT) and complicating ongoing debates on issues like "overthinking" and "test-time scaling." This survey seeks to fill this gap by offering a unified perspective on Long CoT. (1) We first distinguish Long CoT from Short CoT and introduce a novel taxonomy to categorize current reasoning paradigms. (2) Next, we explore the key characteristics of Long CoT: deep reasoning, extensive exploration, and feasible reflection, which enable models to handle more complex tasks and produce more efficient, coherent outcomes compared to the shallower Short CoT. (3) We then investigate key phenomena such as the emergence of Long CoT with these characteristics, including overthinking, and test-time scaling, offering insights into how these processes manifest in practice. (4) Finally, we identify significant research gaps and highlight promising future directions, including the integration of multi-modal reasoning, efficiency improvements, and enhanced knowledge frameworks. By providing a structured overview, this survey aims to inspire future research and further the development of logical reasoning in artificial intelligence.
Authors: Johannes Buchner
Abstract: There exist several techniques for representing the chess board inside the computer. In the first part of this paper, the concepts of the bitboard-representation and the advantages of (rotated) bitboards in move generation are explained. In order to illustrate those ideas practice, the concrete implementation of the move-generator in FUSc# is discussed and we explain a technique how to verify the move-generator with the "perft"-command. We show that the move-generator of FUSc# works 100% correct. The second part of this paper deals with reinforcement learning in computer chess (and beyond). We exemplify the progress that has been made in this field in the last 15-20 years by comparing the "state of the art" from 2002-2008, when FUSc# was developed, with recent innovations connected to "AlphaZero". We discuss how a "FUSc#-Zero" could be implemented and what would be necessary to reduce the number of training games necessary to achieve a good performance. This can be seen as a test case to the general prblem of improving "sample effciency" in reinforcement learning. In the final part, we move beyond computer chess, as the importance of sample effciency extends far beyond board games into a wide range of applications where data is costly, diffcult to obtain, or time consuming to generate. We review some application of the ideas developed in AlphaZero in other domains, i.e. the "other Alphas" like AlphaFold, AlphaTensor, AlphaGeometry and AlphaProof. We also discuss future research and the potential for such methods for ecological economic planning.
Authors: Jianhua Sun, Cewu Lu
Abstract: Reviewing the progress in artificial intelligence over the past decade, various significant advances (e.g. object detection, image generation, large language models) have enabled AI systems to produce more semantically meaningful outputs and achieve widespread adoption in internet scenarios. Nevertheless, AI systems still struggle when it comes to understanding and interacting with the physical world. This reveals an important issue: relying solely on semantic-level concepts learned from internet data (e.g. texts, images) to understand the physical world is far from sufficient -- machine intelligence currently lacks an effective way to learn about the physical world. This research introduces the idea of analytic concept -- representing the concepts related to the physical world through programs of mathematical procedures, providing machine intelligence a portal to perceive, reason about, and interact with the physical world. Except for detailing the design philosophy and providing guidelines for the application of analytic concepts, this research also introduce about the infrastructure that has been built around analytic concepts. I aim for my research to contribute to addressing these questions: What is a proper abstraction of general concepts in the physical world for machine intelligence? How to systematically integrate structured priors with neural networks to constrain AI systems to comply with physical laws?
Authors: Anja Surina, Amin Mansouri, Lars Quaedvlieg, Amal Seddas, Maryna Viazovska, Emmanuel Abbe, Caglar Gulcehre
Abstract: Discovering efficient algorithms for solving complex problems has been an outstanding challenge in mathematics and computer science, requiring substantial human expertise over the years. Recent advancements in evolutionary search with large language models (LLMs) have shown promise in accelerating the discovery of algorithms across various domains, particularly in mathematics and optimization. However, existing approaches treat the LLM as a static generator, missing the opportunity to update the model with the signal obtained from evolutionary exploration. In this work, we propose to augment LLM-based evolutionary search by continuously refining the search operator - the LLM - through reinforcement learning (RL) fine-tuning. Our method leverages evolutionary search as an exploration strategy to discover improved algorithms, while RL optimizes the LLM policy based on these discoveries. Our experiments on three combinatorial optimization tasks - bin packing, traveling salesman, and the flatpack problem - show that combining RL and evolutionary search improves discovery efficiency of improved algorithms, showcasing the potential of RL-enhanced evolutionary strategies to assist computer scientists and mathematicians for more efficient algorithm design.
Authors: Jingyuan Zhang, Qi Wang, Xingguang Ji, Yahui Liu, Yang Yue, Fuzheng Zhang, Di Zhang, Guorui Zhou, Kun Gai
Abstract: Recent advances in automated theorem proving (ATP) through LLMs have highlighted the potential of formal reasoning with Lean 4 codes. However, ATP has not yet be revolutionized by the recent posttraining scaling as demonstrated by Open AI O1/O3 and Deepseek R1. In this work, we investigate the entire posttraining of ATP, aiming to align it with breakthroughs in reasoning models in natural languages. To begin, we continual train current ATP models with a hybrid dataset, which consists of numerous statement-proof pairs, and additional data aimed at incorporating cognitive behaviors that emulate human reasoning and hypothesis refinement. Next, we explore reinforcement learning with the use of outcome reward returned by Lean 4 compiler. Through our designed continual training and reinforcement learning processes, we have successfully improved existing formal provers, including both DeepSeek-Prover-v1.5 and Goedel-Prover, achieving state-of-the-art performance in the field of whole-proof generation. For example, we achieve a 59.8% pass rate (pass@32) on MiniF2F. This is an on-going project and we will progressively update our findings, release our data and training details.
Authors: Andrei Voinea, Robin Kock, Maruf A. Dhali
Abstract: Losing pets can be highly distressing for pet owners, and finding a lost pet is often challenging and time-consuming. An artificial intelligence-based application can significantly improve the speed and accuracy of finding lost pets. To facilitate such an application, this study introduces a contrastive neural network model capable of accurately distinguishing between images of pets. The model was trained on a large dataset of dog images and evaluated through 3-fold cross-validation. Following 350 epochs of training, the model achieved a test accuracy of 90%. Furthermore, overfitting was avoided, as the test accuracy closely matched the training accuracy. Our findings suggest that contrastive neural network models hold promise as a tool for locating lost pets. This paper presents the foundational framework for a potential web application designed to assist users in locating their missing pets. The application will allow users to upload images of their lost pets and provide notifications when matching images are identified within its image database. This functionality aims to enhance the efficiency and accuracy with which pet owners can search for and reunite with their beloved animals.
Authors: Abhishek Ramanathapura Satyanarayana, Maruf A. Dhali
Abstract: Crude oil is an integral component of the world economy and transportation sectors. With the growing demand for crude oil due to its widespread applications, accidental oil spills are unfortunate yet unavoidable. Even though oil spills are difficult to clean up, the first and foremost challenge is to detect them. In this research, the authors test the feasibility of deep encoder-decoder models that can be trained effectively to detect oil spills remotely. The work examines and compares the results from several segmentation models on high dimensional satellite Synthetic Aperture Radar (SAR) image data to pave the way for further in-depth research. Multiple combinations of models are used to run the experiments. The best-performing model is the one with the ResNet-50 encoder and DeepLabV3+ decoder. It achieves a mean Intersection over Union (IoU) of 64.868% and an improved class IoU of 61.549% for the ``oil spill" class when compared with the previous benchmark model, which achieved a mean IoU of 65.05% and a class IoU of 53.38% for the ``oil spill" class.
Authors: Kacper Sokol, Edward Small, Yueqing Xuan
Abstract: Counterfactual explanations are the de facto standard when tasked with interpreting decisions of (opaque) predictive models. Their generation is often subject to technical and domain-specific constraints that aim to maximise their real-life utility. In addition to considering desiderata pertaining to the counterfactual instance itself, guaranteeing existence of a viable path connecting it with the factual data point has recently gained relevance. While current explainability approaches ensure that the steps of such a journey as well as its destination adhere to selected constraints, they neglect the multiplicity of these counterfactual paths. To address this shortcoming we introduce the novel concept of explanatory multiverse that encompasses all the possible counterfactual journeys. We define it using vector spaces, showing how to navigate, reason about and compare the geometry of counterfactual trajectories found within it. To this end, we overview their spatial properties -- such as affinity, branching, divergence and possible future convergence -- and propose an all-in-one metric, called opportunity potential, to quantify them. Notably, the explanatory process offered by our method grants explainees more agency by allowing them to select counterfactuals not only based on their absolute differences but also according to the properties of their connecting paths. To demonstrate real-life flexibility, benefit and efficacy of explanatory multiverse we propose its graph-based implementation, which we use for qualitative and quantitative evaluation on six tabular and image data sets.
Authors: Imad Aouali, Victor-Emmanuel Brunel, David Rohde, Anna Korba
Abstract: In interactive systems, actions are often correlated, presenting an opportunity for more sample-efficient off-policy evaluation (OPE) and learning (OPL) in large action spaces. We introduce a unified Bayesian framework to capture these correlations through structured and informative priors. In this framework, we propose sDM, a generic Bayesian approach for OPE and OPL, grounded in both algorithmic and theoretical foundations. Notably, sDM leverages action correlations without compromising computational efficiency. Moreover, inspired by online Bayesian bandits, we introduce Bayesian metrics that assess the average performance of algorithms across multiple problem instances, deviating from the conventional worst-case assessments. We analyze sDM in OPE and OPL, highlighting the benefits of leveraging action correlations. Empirical evidence showcases the strong performance of sDM.
Authors: Atsuyuki Miyai, Jingkang Yang, Jingyang Zhang, Yifei Ming, Qing Yu, Go Irie, Yixuan Li, Hai Li, Ziwei Liu, Kiyoharu Aizawa
Abstract: This paper introduces a novel task to evaluate the robust understanding capability of Large Multimodal Models (LMMs), termed $\textbf{Unsolvable Problem Detection (UPD)}$. Multiple-choice question answering (MCQA) is widely used to assess the understanding capability of LMMs, but it does not guarantee that LMMs truly comprehend the answer. UPD assesses the LMM's ability to withhold answers when encountering unsolvable problems of MCQA, verifying whether the model truly understands the answer. UPD encompasses three problems: Absent Answer Detection (AAD), Incompatible Answer Set Detection (IASD), and Incompatible Visual Question Detection (IVQD), covering unsolvable cases like answer-lacking or incompatible choices and image-question mismatches. For the evaluation, we introduce the MM-UPD Bench, a benchmark for assessing performance across various ability dimensions. Our experiments reveal that even most LMMs, which demonstrate adequate performance on existing benchmarks, struggle significantly with MM-UPD, underscoring a novel aspect of trustworthiness that current benchmarks have overlooked. A detailed analysis shows that LMMs have different bottlenecks and chain-of-thought and self-reflection improved performance for LMMs with the bottleneck in their LLM capability. We hope our insights will enhance the broader understanding and development of more reliable LMMs.
Authors: Fanxu Meng, Zhaohui Wang, Muhan Zhang
Abstract: To parameter-efficiently fine-tune (PEFT) large language models (LLMs), the low-rank adaptation (LoRA) method approximates the model changes $\Delta W \in \mathbb{R}^{m \times n}$ through the product of two matrices $A \in \mathbb{R}^{m \times r}$ and $B \in \mathbb{R}^{r \times n}$, where $r \ll \min(m, n)$, $A$ is initialized with Gaussian noise, and $B$ with zeros. LoRA freezes the original model $W$ and updates the "Noise & Zero" adapter, which may lead to slow convergence. To overcome this limitation, we introduce Principal Singular values and Singular vectors Adaptation (PiSSA). PiSSA shares the same architecture as LoRA, but initializes the adaptor matrices $A$ and $B$ with the principal components of the original matrix $W$, and put the remaining components into a residual matrix $W^{res} \in \mathbb{R}^{m \times n}$ which is frozen during fine-tuning. Compared to LoRA, PiSSA updates the principal components while freezing the "residual" parts, allowing faster convergence and enhanced performance. Comparative experiments of PiSSA and LoRA across 12 different models, ranging from 184M to 70B, encompassing 5 NLG and 8 NLU tasks, reveal that PiSSA consistently outperforms LoRA under identical experimental setups. On the GSM8K benchmark, Mistral-7B fine-tuned with PiSSA achieves an accuracy of 72.86%, surpassing LoRA's 67.7% by 5.16%. Due to the same architecture, PiSSA is also compatible with quantization to further reduce the memory requirement of fine-tuning. Compared to QLoRA, QPiSSA exhibits smaller quantization errors in the initial stages. Fine-tuning LLaMA-3-70B on GSM8K, QPiSSA attains an accuracy of 86.05%, exceeding the performances of QLoRA at 81.73%. Leveraging a fast SVD technique, PiSSA can be initialized in only a few seconds, presenting a negligible cost for transitioning from LoRA to PiSSA. Code is available at https://github.com/GraphPKU/PiSSA.
Authors: Simone Barandoni, Filippo Chiarello, Lorenzo Cascone, Emiliano Marrale, Salvatore Puccio
Abstract: In the rapidly evolving landscape of Natural Language Processing (NLP), Large Language Models (LLMs) have emerged as powerful tools for many tasks, such as extracting valuable insights from vast amounts of textual data. In this study, we conduct a comparative analysis of LLMs for the extraction of travel customer needs from TripAdvisor and Reddit posts. Leveraging a diverse range of models, including both open-source and proprietary ones such as GPT-4 and Gemini, we aim to elucidate their strengths and weaknesses in this specialized domain. Through an evaluation process involving metrics such as BERTScore, ROUGE, and BLEU, we assess the performance of each model in accurately identifying and summarizing customer needs. Our findings highlight the efficacy of opensource LLMs, particularly Mistral 7B, in achieving comparable performance to larger closed models while offering affordability and customization benefits. Additionally, we underscore the importance of considering factors such as model size, resource requirements, and performance metrics when selecting the most suitable LLM for customer needs analysis tasks. Overall, this study contributes valuable insights for businesses seeking to leverage advanced NLP techniques to enhance customer experience and drive operational efficiency in the travel industry.
Authors: Sanghyeon Na, Yonggyu Kim, Hyunjoon Lee
Abstract: Human image generation is a key focus in image synthesis due to its broad applications, but even slight inaccuracies in anatomy, pose, or details can compromise realism. To address these challenges, we explore Direct Preference Optimization (DPO), which trains models to generate preferred (winning) images while diverging from non-preferred (losing) ones. However, conventional DPO methods use generated images as winning images, limiting realism. To overcome this limitation, we propose an enhanced DPO approach that incorporates high-quality real images as winning images, encouraging outputs to resemble real images rather than generated ones. However, implementing this concept is not a trivial task. Therefore, our approach, HG-DPO (Human image Generation through DPO), employs a novel curriculum learning framework that gradually improves the output of the model toward greater realism, making training more feasible. Furthermore, HG-DPO effectively adapts to personalized text-to-image tasks, generating high-quality and identity-specific images, which highlights the practical value of our approach.
Authors: Hanyang Zhong, Liman Wang, Wenting Cao, Zeyuan Sun
Abstract: This paper examines the role of cognitive biases in the decision-making processes of large language models (LLMs), challenging the conventional goal of eliminating all biases. When properly balanced, we show that certain cognitive biases can enhance decision-making efficiency through rational deviations and heuristic shortcuts. By introducing heuristic moderation and an abstention option, which allows LLMs to withhold responses when uncertain, we reduce error rates, improve decision accuracy, and optimize decision rates. Using the Balance Rigor and Utility (BRU) dataset, developed through expert collaboration, our findings demonstrate that targeted inspection of cognitive biases aligns LLM decisions more closely with human reasoning, enhancing reliability and suggesting strategies for future improvements. This approach offers a novel way to leverage cognitive biases to improve the practical utility of LLMs across various applications.
Authors: Yuni Susanti, Nina Holsmoelle
Abstract: This study explores the capability of Large Language Models (LLMs) to evaluate causality in causal graphs generated by conventional statistical causal discovery methods-a task traditionally reliant on manual assessment by human subject matter experts. To bridge this gap in causality assessment, LLMs are employed to evaluate the causal relationships by determining whether a causal connection between variable pairs can be inferred from textual context. Our study compares two approaches: (1) prompting-based method for zero-shot and few-shot causal inference and, (2) fine-tuning language models for the causal relation prediction task. While prompt-based LLMs have demonstrated versatility across various NLP tasks, our experiments on biomedical and general-domain datasets show that fine-tuned models consistently outperform them, achieving up to a 20.5-point improvement in F1 score-even when using smaller-parameter language models. These findings provide valuable insights into the strengths and limitations of both approaches for causal graph evaluation.
Authors: Silin Meng, Yiwei Wang, Cheng-Fu Yang, Nanyun Peng, Kai-Wei Chang
Abstract: Path planning is a fundamental scientific problem in robotics and autonomous navigation, requiring the derivation of efficient routes from starting to destination points while avoiding obstacles. Traditional algorithms like A* and its variants are capable of ensuring path validity but suffer from significant computational and memory inefficiencies as the state space grows. Conversely, large language models (LLMs) excel in broader environmental analysis through contextual understanding, providing global insights into environments. However, they fall short in detailed spatial and temporal reasoning, often leading to invalid or inefficient routes. In this work, we propose LLM-A*, an new LLM based route planning method that synergistically combines the precise pathfinding capabilities of A* with the global reasoning capability of LLMs. This hybrid approach aims to enhance pathfinding efficiency in terms of time and space complexity while maintaining the integrity of path validity, especially in large-scale scenarios. By integrating the strengths of both methodologies, LLM-A* addresses the computational and memory limitations of conventional algorithms without compromising on the validity required for effective pathfinding.
Authors: Irene Siragusa, Salvatore Contino, Massimo La Ciura, Rosario Alicata, Roberto Pirrone
Abstract: The increasing interest in developing Artificial Intelligence applications in the medical domain, suffers from the lack of high-quality data set, mainly due to privacy-related issues. In addition, the recent increase in Vision Language Models (VLM) leads to the need for multimodal medical data sets, where clinical reports and findings are attached to the corresponding medical scans. This paper illustrates the entire workflow for building the MedPix 2.0 data set. Starting with the well-known multimodal data set MedPix, mainly used by physicians, nurses, and healthcare students for Continuing Medical Education purposes, a semi-automatic pipeline was developed to extract visual and textual data followed by a manual curing procedure in which noisy samples were removed, thus creating a MongoDB database. Along with the data set, we developed a Graphical User Interface aimed at navigating efficiently the MongoDB instance and obtaining the raw data that can be easily used for training and/or fine-tuning VLMs. To enforce this point, in this work, we first recall DR-Minerva, a Retrieve Augmented Generation-based VLM model trained upon MedPix 2.0. DR-Minerva predicts the body part and the modality used to scan its input image. We also propose the extension of DR-Minerva with a Knowledge Graph that uses Llama 3.1 Instruct 8B, and leverages MedPix 2.0. The resulting architecture can be queried in a end-to-end manner, as a medical decision support system. MedPix 2.0 is available on GitHub https://github.com/CHILab1/MedPix-2.0
Authors: Yilie Huang, Yanwei Jia, Xun Yu Zhou
Abstract: We study reinforcement learning (RL) for a class of continuous-time linear-quadratic (LQ) control problems for diffusions, where states are scalar-valued and running control rewards are absent but volatilities of the state processes depend on both state and control variables. We apply a model-free approach that relies neither on knowledge of model parameters nor on their estimations, and devise an RL algorithm to learn the optimal policy parameter directly. Our main contributions include the introduction of an exploration schedule and a regret analysis of the proposed algorithm. We provide the convergence rate of the policy parameter to the optimal one, and prove that the algorithm achieves a regret bound of $O(N^{\frac{3}{4}})$ up to a logarithmic factor, where $N$ is the number of learning episodes. We conduct a simulation study to validate the theoretical results and demonstrate the effectiveness and reliability of the proposed algorithm. We also perform numerical comparisons between our method and those of the recent model-based stochastic LQ RL studies adapted to the state- and control-dependent volatility setting, demonstrating a better performance of the former in terms of regret bounds.
Authors: Xingyue Lin, Xingjian Hu, Shuai Peng, Jianhua Zhu, Liangcai Gao
Abstract: Sketching is a powerful artistic technique for capturing essential visual information about real-world objects and has increasingly attracted attention in image synthesis research. However, the field lacks a unified benchmark to evaluate the performance of various synthesis methods. To address this, we propose SketchRef, the first comprehensive multi-task evaluation benchmark for sketch synthesis. SketchRef fully leverages the shared characteristics between sketches and reference photos. It introduces two primary tasks: category prediction and structural consistency estimation, the latter being largely overlooked in previous studies. These tasks are further divided into five sub-tasks across four domains: animals, common things, human body, and faces. Recognizing the inherent trade-off between recognizability and simplicity in sketches, we are the first to quantify this balance by introducing a recognizability calculation method constrained by simplicity, mRS, ensuring fair and meaningful evaluations. To validate our approach, we collected 7,920 responses from art enthusiasts, confirming the effectiveness of our proposed evaluation metrics. Additionally, we evaluate the performance of existing sketch synthesis methods on our benchmark, highlighting their strengths and weaknesses. We hope this study establishes a standardized benchmark and offers valuable insights for advancing sketch synthesis algorithms.
Authors: Yang Nan, Huichi Zhou, Xiaodan Xing, Guang Yang
Abstract: Recent advancements in Large Vision-Language Models (LVLMs) have demonstrated remarkable capabilities across diverse tasks, garnering significant attention in AI communities. However, their performance and reliability in specialized domains such as medicine remain insufficiently assessed. In particular, most assessments over-concentrate on evaluating VLMs based on simple Visual Question Answering (VQA) on multi-modality data, while ignoring the in-depth characteristics of LVLMs. In this study, we introduce RadVUQA, a novel Radiological Visual Understanding and Question Answering benchmark, to comprehensively evaluate existing LVLMs. RadVUQA mainly validates LVLMs across five dimensions: 1) Anatomical understanding, assessing the models' ability to visually identify biological structures; 2) Multimodal comprehension, which involves the capability of interpreting linguistic and visual instructions to produce desired outcomes; 3) Quantitative and spatial reasoning, evaluating the models' spatial awareness and proficiency in combining quantitative analysis with visual and linguistic information; 4) Physiological knowledge, measuring the models' capability to comprehend functions and mechanisms of organs and systems; and 5) Robustness, which assesses the models' capabilities against unharmonized and synthetic data. The results indicate that both generalized LVLMs and medical-specific LVLMs have critical deficiencies with weak multimodal comprehension and quantitative reasoning capabilities. Our findings reveal the large gap between existing LVLMs and clinicians, highlighting the urgent need for more robust and intelligent LVLMs. The code is available at https://github.com/Nandayang/RadVUQA
Authors: Hongjin Qian, Zheng Liu, Peitian Zhang, Kelong Mao, Defu Lian, Zhicheng Dou, Tiejun Huang
Abstract: Processing long contexts presents a significant challenge for large language models (LLMs). While recent advancements allow LLMs to handle much longer contexts than before (e.g., 32K or 128K tokens), it is computationally expensive and can still be insufficient for many applications. Retrieval-Augmented Generation (RAG) is considered a promising strategy to address this problem. However, conventional RAG methods face inherent limitations because of two underlying requirements: 1) explicitly stated queries, and 2) well-structured knowledge. These conditions, however, do not hold in general long-context processing tasks. In this work, we propose MemoRAG, a novel RAG framework empowered by global memory-augmented retrieval. MemoRAG features a dual-system architecture. First, it employs a light but long-range system to create a global memory of the long context. Once a task is presented, it generates draft answers, providing useful clues for the retrieval tools to locate relevant information within the long context. Second, it leverages an expensive but expressive system, which generates the final answer based on the retrieved information. Building upon this fundamental framework, we realize the memory module in the form of KV compression, and reinforce its memorization and cluing capacity from the Generation quality's Feedback (a.k.a. RLGF). In our experiments, MemoRAG achieves superior performances across a variety of long-context evaluation tasks, not only complex scenarios where traditional RAG methods struggle, but also simpler ones where RAG is typically applied.
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 and 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 not available at every forecast cycle. We develop a deep-learning algorithm called GeoCenter; besides a few scalars in the operational ATCF, it relies only on geostationary 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: Qingzhao Zhang, Ziyang Xiong, Z. Morley Mao
Abstract: Safety is a paramount concern for large language models (LLMs) in open deployment, motivating the development of safeguard methods that enforce ethical and responsible use through safety alignment or guardrail mechanisms. Jailbreak attacks that exploit the \emph{false negatives} of safeguard methods have emerged as a prominent research focus in the field of LLM security. However, we found that the malicious attackers could also exploit false positives of safeguards, i.e., fooling the safeguard model to block safe content mistakenly, leading to a denial-of-service (DoS) affecting LLM users. To bridge the knowledge gap of this overlooked threat, we explore multiple attack methods that include inserting a short adversarial prompt into user prompt templates and corrupting the LLM on the server by poisoned fine-tuning. In both ways, the attack triggers safeguard rejections of user requests from the client. Our evaluation demonstrates the severity of this threat across multiple scenarios. For instance, in the scenario of white-box adversarial prompt injection, the attacker can use our optimization process to automatically generate seemingly safe adversarial prompts, approximately only 30 characters long, that universally block over 97% of user requests on Llama Guard 3. These findings reveal a new dimension in LLM safeguard evaluation -- adversarial robustness to false positives.
Authors: Tommaso Giorgi, Lorenzo Cima, Tiziano Fagni, Marco Avvenuti, Stefano Cresci
Abstract: The rise of online platforms exacerbated the spread of hate speech, demanding scalable and effective detection. However, the accuracy of hate speech detection systems heavily relies on human-labeled data, which is inherently susceptible to biases. While previous work has examined the issue, the interplay between the characteristics of the annotator and those of the target of the hate are still unexplored. We fill this gap by leveraging an extensive dataset with rich socio-demographic information of both annotators and targets, uncovering how human biases manifest in relation to the target's attributes. Our analysis surfaces the presence of widespread biases, which we quantitatively describe and characterize based on their intensity and prevalence, revealing marked differences. Furthermore, we compare human biases with those exhibited by persona-based LLMs. Our findings indicate that while persona-based LLMs do exhibit biases, these differ significantly from those of human annotators. Overall, our work offers new and nuanced results on human biases in hate speech annotations, as well as fresh insights into the design of AI-driven hate speech detection systems.
Authors: Zhiyuan Peng, Jinming Nian, Alexandre Evfimievski, Yi Fang
Abstract: Large Language Models (LLMs) are widely used in Conversational AI systems to generate responses to user inquiries. However, many natural questions lack well-defined answers. While existing studies primarily focus on question types such as false premises, they often overlook out-of-scope questions, where the provided document is semantically highly similar to the query but does not contain the required answer. In this paper, we propose a guided hallucination-based method to efficiently generate a diverse set of out-of-scope questions from a given document corpus. We then evaluate multiple LLMs based on their effectiveness in confusion detection and appropriate response generation. Furthermore, we introduce an improved method for detecting such out-of-scope questions, enhancing the reliability of LLM-based question-answering systems.
Authors: Andrew Liu, Axel Elaldi, Nathan Russell, Olivia Viessmann
Abstract: Efficient encoding and representation of large 3D molecular structures with high fidelity is critical for biomolecular design applications. Despite this, many representation learning approaches restrict themselves to modeling smaller systems or use coarse-grained approximations of the systems, for example modeling proteins at the resolution of amino acid residues rather than at the level of individual atoms. To address this, we develop quantized auto-encoders that learn atom-level tokenizations of complete proteins, RNA and small molecule structures with reconstruction accuracies well below 1 Angstrom. We demonstrate that a simple Mamba state space model architecture is efficient compared to an SE(3)-invariant IPA architecture, reaches competitive accuracies and can scale to systems with almost 100,000 atoms. The learned structure tokens of bio2token may serve as the input for all-atom generative models in the future.
Authors: Yihua Shao, Yan Gu, Siyu Chen, Haiyang Liu, Zijian Ling, Minxi Yan, Ziyang Yan, Chenyu Zhang, Michele Magno, Haotong Qin, Yan Wang, Jingcai Guo, Ling Shao, Hao Tang
Abstract: Large language models (LLMs) show impressive performance in solving complex language tasks. However, its large number of parameters presents significant challenges for the deployment. So, compressing LLMs to low bits can enable to deploy on resource-constrained devices. To address this problem, we propose gradient-aware weight quantization (GWQ), the first quantization approach for low-bit weight quantization that leverages gradients to localize outliers, requiring only a minimal amount of calibration data for outlier detection. GWQ retains the top 1\% outliers preferentially at FP16 precision, while the remaining non-outlier weights are stored in a low-bit. We widely evaluate GWQ on different task include language modeling, grounding detection, massive multitask language understanding and vision-language question and answering. Results show that models quantified by GWQ performs better than other quantization method. During quantization process, GWQ only need one calibration set to realize effective quant. Also, GWQ achieves 1.2x inference speedup in comparison to the original model and effectively reduces the inference memory.
Authors: Ting-Ju Wei, Chuin-Shan Chen
Abstract: The rapid advancement of machine learning has unlocked numerous opportunities for materials science, particularly in accelerating the design and analysis of materials. However, a significant challenge lies in the scarcity and high cost of obtaining high-quality materials datasets. While foundation models pre-trained on large datasets have excelled in fields like natural language processing by leveraging latent features through transfer learning, their application in materials science remains limited. Here, we present a foundation model specifically designed for composite materials. Pre-trained on a dataset of short-fiber composites to learn robust latent features, the model accurately predicts homogenized stiffness during transfer learning, even with limited training data. Additionally, our model effectively predicts the material's nonlinear behavior by transferring these learned features to an Interaction-based Material Network, which is a constitutive surrogate model. These results demonstrate the potential of our foundation model to capture complex material behaviors. Our findings validate the feasibility and effectiveness of foundation models in composite materials. We anticipate extending this approach to more complex three-dimensional composite materials, polycrystalline materials, and beyond. Moreover, this framework enables high-accuracy predictions even when experimental data are scarce, paving the way for more efficient and cost-effective materials design and analysis.
Authors: Meher Bhardwaj, Hrishikesh Ethari, Dennis Singh Moirangthem
Abstract: The SQL-to-text generation task traditionally uses template base, Seq2Seq, tree-to-sequence, and graph-to-sequence models. Recent models take advantage of pre-trained generative language models for this task in the Seq2Seq framework. However, treating SQL as a sequence of inputs to the pre-trained models is not optimal. In this work, we put forward a new SQL intermediate representation called EzSQL to align SQL with the natural language text sequence. EzSQL simplifies the SQL queries and brings them closer to natural language text by modifying operators and keywords, which can usually be described in natural language. EzSQL also removes the need for set operators. Our proposed SQL-to-text generation model uses EzSQL as the input to a pre-trained generative language model for generating the text descriptions. We demonstrate that our model is an effective state-of-the-art method to generate text narrations from SQL queries on the WikiSQL and Spider datasets. We also show that by generating pretraining data using our SQL-to-text generation model, we can enhance the performance of Text-to-SQL parsers.
Authors: Yoshua Bengio, S\"oren Mindermann, Daniel Privitera, Tamay Besiroglu, Rishi Bommasani, Stephen Casper, Yejin Choi, Danielle Goldfarb, Hoda Heidari, Leila Khalatbari, Shayne Longpre, Vasilios Mavroudis, Mantas Mazeika, Kwan Yee Ng, Chinasa T. Okolo, Deborah Raji, Theodora Skeadas, Florian Tram\`er, Bayo Adekanmbi, Paul Christiano, David Dalrymple, Thomas G. Dietterich, Edward Felten, Pascale Fung, Pierre-Olivier Gourinchas, Nick Jennings, Andreas Krause, Percy Liang, Teresa Ludermir, Vidushi Marda, Helen Margetts, John A. McDermid, Arvind Narayanan, Alondra Nelson, Alice Oh, Gopal Ramchurn, Stuart Russell, Marietje Schaake, Dawn Song, Alvaro Soto, Lee Tiedrich, Ga\"el Varoquaux, Andrew Yao, Ya-Qin Zhang
Abstract: This is the interim publication of the first International Scientific Report on the Safety of Advanced AI. The report synthesises the scientific understanding of general-purpose AI -- AI that can perform a wide variety of tasks -- with a focus on understanding and managing its risks. A diverse group of 75 AI experts contributed to this report, including an international Expert Advisory Panel nominated by 30 countries, the EU, and the UN. Led by the Chair, these independent experts collectively had full discretion over the report's content. The final report is available at arXiv:2501.17805
Authors: Pan Wang, Qiang Zhou, Yawen Wu, Tianlong Chen, Jingtong Hu
Abstract: Multimodal Sentiment Analysis (MSA) leverages heterogeneous modalities, such as language, vision, and audio, to enhance the understanding of human sentiment. While existing models often focus on extracting shared information across modalities or directly fusing heterogeneous modalities, such approaches can introduce redundancy and conflicts due to equal treatment of all modalities and the mutual transfer of information between modality pairs. To address these issues, we propose a Disentangled-Language-Focused (DLF) multimodal representation learning framework, which incorporates a feature disentanglement module to separate modality-shared and modality-specific information. To further reduce redundancy and enhance language-targeted features, four geometric measures are introduced to refine the disentanglement process. A Language-Focused Attractor (LFA) is further developed to strengthen language representation by leveraging complementary modality-specific information through a language-guided cross-attention mechanism. The framework also employs hierarchical predictions to improve overall accuracy. Extensive experiments on two popular MSA datasets, CMU-MOSI and CMU-MOSEI, demonstrate the significant performance gains achieved by the proposed DLF framework. Comprehensive ablation studies further validate the effectiveness of the feature disentanglement module, language-focused attractor, and hierarchical predictions. Our code is available at https://github.com/pwang322/DLF.
Authors: Yuxi Sun, Wei Gao, Jing Ma, Hongzhan Lin, Ziyang Luo, Wenxuan Zhang
Abstract: With the rise and widespread use of Large Language Models (LLMs), ensuring their safety is crucial to prevent harm to humans and promote ethical behaviors. However, directly assessing value valence (i.e., support or oppose) by leveraging large-scale data training is untrustworthy and inexplainable. We assume that emulating humans to rely on social norms to make moral decisions can help LLMs understand and predict moral judgment. However, capturing human values remains a challenge, as multiple related norms might conflict in specific contexts. Consider norms that are upheld by the majority and promote the well-being of society are more likely to be accepted and widely adopted (e.g., "don't cheat,"). Therefore, it is essential for LLM to identify the appropriate norms for a given scenario before making moral decisions. To this end, we introduce a novel moral judgment approach called \textit{ClarityEthic} that leverages LLMs' reasoning ability and contrastive learning to uncover relevant social norms for human actions from different perspectives and select the most reliable one to enhance judgment accuracy. Extensive experiments demonstrate that our method outperforms state-of-the-art approaches in moral judgment tasks. Moreover, human evaluations confirm that the generated social norms provide plausible explanations that support the judgments. This suggests that modeling human moral judgment with the emulating humans moral strategy is promising for improving the ethical behaviors of LLMs.
Authors: Burak Ekim, Girmaw Abebe Tadesse, Caleb Robinson, Gilles Hacheme, Michael Schmitt, Rahul Dodhia, Juan M. Lavista Ferres
Abstract: Training robust deep learning models is crucial in Earth Observation, where globally deployed models often face distribution shifts that degrade performance, especially in low-data regions. Out-of-distribution (OOD) detection addresses this by identifying inputs that deviate from in-distribution (ID) data. However, existing methods either assume access to OOD data or compromise primary task performance, limiting real-world use. We introduce TARDIS, a post-hoc OOD detection method designed for scalable geospatial deployment. Our core innovation lies in generating surrogate distribution labels by leveraging ID data within the feature space. TARDIS takes a pre-trained model, ID data, and data from an unknown distribution (WILD), separates WILD into surrogate ID and OOD labels based on internal activations, and trains a binary classifier to detect distribution shifts. We validate on EuroSAT and xBD across 17 setups covering covariate and semantic shifts, showing near-upper-bound surrogate labeling performance in 13 cases and matching the performance of top post-hoc activation- and scoring-based methods. Finally, deploying TARDIS on Fields of the World reveals actionable insights into pre-trained model behavior at scale. The code is available at \href{https://github.com/microsoft/geospatial-ood-detection}{https://github.com/microsoft/geospatial-ood-detection}
URLs: https://github.com/microsoft/geospatial-ood-detection, https://github.com/microsoft/geospatial-ood-detection
Authors: Chulun Zhou, Qiujing Wang, Mo Yu, Xiaoqian Yue, Rui Lu, Jiangnan Li, Yifan Zhou, Shunchi Zhang, Jie Zhou, Wai Lam
Abstract: Theory-of-Mind (ToM) is a fundamental psychological capability that allows humans to understand and interpret the mental states of others. Humans infer others' thoughts by integrating causal cues and indirect clues from broad contextual information, often derived from past interactions. In other words, human ToM heavily relies on the understanding about the backgrounds and life stories of others. Unfortunately, this aspect is largely overlooked in existing benchmarks for evaluating machines' ToM capabilities, due to their usage of short narratives without global context, especially personal background of characters. In this paper, we verify the importance of comprehensive contextual understanding about personal backgrounds in ToM and assess the performance of LLMs in such complex scenarios. To achieve this, we introduce CharToM benchmark, comprising 1,035 ToM questions based on characters from classic novels. Our human study reveals a significant disparity in performance: the same group of educated participants performs dramatically better when they have read the novels compared to when they have not. In parallel, our experiments on state-of-the-art LLMs, including the very recent o1 and DeepSeek-R1 models, show that LLMs still perform notably worse than humans, despite that they have seen these stories during pre-training. This highlights the limitations of current LLMs in capturing the nuanced contextual information required for ToM reasoning.
Authors: Mingyang Song, Zhaochen Su, Xiaoye Qu, Jiawei Zhou, Yu Cheng
Abstract: Process-level Reward Models (PRMs) are crucial for complex reasoning and decision-making tasks, where each intermediate step plays an important role in the reasoning process. Since language models are prone to various types of errors during the reasoning process, PRMs are required to possess nuanced capabilities for detecting various implicit error types in real-world scenarios. However, current benchmarks primarily focus on step correctness, failing to evaluate PRMs' performance systematically. To address this gap, we introduce PRMBench, a process-level benchmark specifically designed to assess the fine-grained error detection capabilities of PRMs. PRMBench comprises 6,216 carefully designed problems and 83,456 step-level labels, evaluating models across multiple dimensions, including simplicity, soundness, and sensitivity. In our experiments on 15 models, spanning both open-source PRMs and closed-source large language models prompted as critic models, we uncover significant weaknesses in current PRMs. These findings underscore the challenges inherent in process-level evaluation and highlight key directions for future research. We hope PRMBench can be a robust bench for advancing research on PRM evaluation and development.
Authors: Yuhui Zhang, Yuchang Su, Yiming Liu, Xiaohan Wang, James Burgess, Elaine Sui, Chenyu Wang, Josiah Aklilu, Alejandro Lozano, Anjiang Wei, Ludwig Schmidt, Serena Yeung-Levy
Abstract: The rapid development of vision language models (VLMs) demands rigorous and reliable evaluation. However, current visual question answering (VQA) benchmarks often depend on open-ended questions, making accurate evaluation difficult due to the variability in natural language responses. To address this, we introduce AutoConverter, an agentic framework that automatically converts these open-ended questions into multiple-choice format, enabling objective evaluation while reducing the costly multiple-choice question creation process. Our experiments demonstrate that AutoConverter can generate correct and challenging multiple-choice questions, with VLMs demonstrating consistently similar or lower accuracy on these questions compared to human-created ones. Using AutoConverter, we construct VMCBench, a benchmark created by transforming 20 existing VQA datasets into a unified multiple-choice format, totaling 9,018 questions. We comprehensively evaluate 33 state-of-the-art VLMs on VMCBench, setting a new standard for scalable, consistent, and reproducible VLM evaluation.
Authors: Peiyan Yue, Die Cai, Chu Guo, Mengxing Liu, Jun Xia, Yi Wang
Abstract: Accurate automated segmentation of tibial plateau fractures (TPF) from computed tomography (CT) requires large amounts of annotated data to train deep learning models, but obtaining such annotations presents unique challenges. The process demands expert knowledge to identify diverse fracture patterns, assess severity, and account for individual anatomical variations, making the annotation process highly time-consuming and expensive. Although semi-supervised learning methods can utilize unlabeled data, existing approaches often struggle with the complexity and variability of fracture morphologies, as well as limited generalizability across datasets. To tackle these issues, we propose an effective training strategy based on masked autoencoder (MAE) for the accurate TPF segmentation in CT. Our method leverages MAE pretraining to capture global skeletal structures and fine-grained fracture details from unlabeled data, followed by fine-tuning with a small set of labeled data. This strategy reduces the dependence on extensive annotations while enhancing the model's ability to learn generalizable and transferable features. The proposed method is evaluated on an in-house dataset containing 180 CT scans with TPF. Experimental results demonstrate that our method consistently outperforms semi-supervised methods, achieving an average Dice similarity coefficient (DSC) of 95.81%, average symmetric surface distance (ASSD) of 1.91mm, and Hausdorff distance (95HD) of 9.42mm with only 20 annotated cases. Moreover, our method exhibits strong transferability when applying to another public pelvic CT dataset with hip fractures, highlighting its potential for broader applications in fracture segmentation tasks.
Authors: Bucher Sahyouni, Matthew Vowels, Liqun Chen, Simon Hadfield
Abstract: The development of fair and unbiased machine learning models remains an ongoing objective for researchers in the field of artificial intelligence. We introduce the Differential Adjusted Parity (DAP) loss to produce unbiased informative representations. It utilises a differentiable variant of the adjusted parity metric to create a unified objective function. By combining downstream task classification accuracy and its inconsistency across sensitive feature domains, it provides a single tool to increase performance and mitigate bias. A key element in this approach is the use of soft balanced accuracies. In contrast to previous non-adversarial approaches, DAP does not suffer a degeneracy where the metric is satisfied by performing equally poorly across all sensitive domains. It outperforms several adversarial models on downstream task accuracy and fairness in our analysis. Specifically, it improves the demographic parity, equalized odds and sensitive feature accuracy by as much as 22.5\%, 44.1\% and 40.1\%, respectively, when compared to the best performing adversarial approaches on these metrics. Overall, the DAP loss and its associated metric can play a significant role in creating more fair machine learning models.
Authors: Ebenezer Tarubinga, Jenifer Kalafatovich, Seong-Whan Lee
Abstract: Semi-supervised semantic segmentation (SSSS) aims to improve segmentation performance by utilizing large amounts of unlabeled data with limited labeled samples. Existing methods often suffer from coupling, where over-reliance on initial labeled data leads to suboptimal learning; confirmation bias, where incorrect predictions reinforce themselves repeatedly; and boundary blur caused by limited boundary-awareness and ambiguous edge cues. To address these issues, we propose CW-BASS, a novel framework for SSSS. In order to mitigate the impact of incorrect predictions, we assign confidence weights to pseudo-labels. Additionally, we leverage boundary-delineation techniques, which, despite being extensively explored in weakly-supervised semantic segmentation (WSSS), remain underutilized in SSSS. Specifically, our method: (1) reduces coupling via a confidence-weighted loss that adjusts pseudo-label influence based on their predicted confidence scores, (2) mitigates confirmation bias with a dynamic thresholding mechanism that learns to filter out pseudo-labels based on model performance, (3) tackles boundary blur using a boundary-aware module to refine segmentation near object edges, and (4) reduces label noise through a confidence decay strategy that progressively refines pseudo-labels during training. Extensive experiments on Pascal VOC 2012 and Cityscapes demonstrate that CW-BASS achieves state-of-the-art performance. Notably, CW-BASS achieves a 65.9% mIoU on Cityscapes under a challenging and underexplored 1/30 (3.3%) split (100 images), highlighting its effectiveness in limited-label settings. Our code is available at https://github.com/psychofict/CW-BASS.
Authors: Nikita Shvetsov, Thomas K. Kilvaer, Masoud Tafavvoghi, Anders Sildnes, Kajsa M{\o}llersen, Lill-Tove Rasmussen Busund, Lars Ailo Bongo
Abstract: Developing clinically useful cell-level analysis tools in digital pathology remains challenging due to limitations in dataset granularity, inconsistent annotations, high computational demands, and difficulties integrating new technologies into workflows. To address these issues, we propose a solution that enhances data quality, model performance, and usability by creating a lightweight, extensible cell segmentation and classification model. First, we update data labels through cross-relabeling to refine annotations of PanNuke and MoNuSAC, producing a unified dataset with seven distinct cell types. Second, we leverage the H-Optimus foundation model as a fixed encoder to improve feature representation for simultaneous segmentation and classification tasks. Third, to address foundation models' computational demands, we distill knowledge to reduce model size and complexity while maintaining comparable performance. Finally, we integrate the distilled model into QuPath, a widely used open-source digital pathology platform. Results demonstrate improved segmentation and classification performance using the H-Optimus-based model compared to a CNN-based model. Specifically, average $R^2$ improved from 0.575 to 0.871, and average $PQ$ score improved from 0.450 to 0.492, indicating better alignment with actual cell counts and enhanced segmentation quality. The distilled model maintains comparable performance while reducing parameter count by a factor of 48. By reducing computational complexity and integrating into workflows, this approach may significantly impact diagnostics, reduce pathologist workload, and improve outcomes. Although the method shows promise, extensive validation is necessary prior to clinical deployment.
Authors: Yuxuan Bian, Zhaoyang Zhang, Xuan Ju, Mingdeng Cao, Liangbin Xie, Ying Shan, Qiang Xu
Abstract: Video inpainting, which aims to restore corrupted video content, has experienced substantial progress. Despite these advances, existing methods, whether propagating unmasked region pixels through optical flow and receptive field priors, or extending image-inpainting models temporally, face challenges in generating fully masked objects or balancing the competing objectives of background context preservation and foreground generation in one model, respectively. To address these limitations, we propose a novel dual-stream paradigm VideoPainter that incorporates an efficient context encoder (comprising only 6% of the backbone parameters) to process masked videos and inject backbone-aware background contextual cues to any pre-trained video DiT, producing semantically consistent content in a plug-and-play manner. This architectural separation significantly reduces the model's learning complexity while enabling nuanced integration of crucial background context. We also introduce a novel target region ID resampling technique that enables any-length video inpainting, greatly enhancing our practical applicability. Additionally, we establish a scalable dataset pipeline leveraging current vision understanding models, contributing VPData and VPBench to facilitate segmentation-based inpainting training and assessment, the largest video inpainting dataset and benchmark to date with over 390K diverse clips. Using inpainting as a pipeline basis, we also explore downstream applications including video editing and video editing pair data generation, demonstrating competitive performance and significant practical potential. Extensive experiments demonstrate VideoPainter's superior performance in both any-length video inpainting and editing, across eight key metrics, including video quality, mask region preservation, and textual coherence.
Authors: Xinjie Liu, Cyrus Neary, Kushagra Gupta, Christian Ellis, Ufuk Topcu, David Fridovich-Keil
Abstract: Many reinforcement learning (RL) algorithms require large amounts of data, prohibiting their use in applications where frequent interactions with operational systems are infeasible, or high-fidelity simulations are expensive or unavailable. Meanwhile, low-fidelity simulators--such as reduced-order models, heuristic reward functions, or generative world models--can cheaply provide useful data for RL training, even if they are too coarse for direct sim-to-real transfer. We propose multi-fidelity policy gradients (MFPGs), an RL framework that mixes a small amount of data from the target environment with a large volume of low-fidelity simulation data to form unbiased, reduced-variance estimators (control variates) for on-policy policy gradients. We instantiate the framework by developing multi-fidelity variants of two policy gradient algorithms: REINFORCE and proximal policy optimization. Experimental results across a suite of simulated robotics benchmark problems demonstrate that when target-environment samples are limited, MFPG achieves up to 3.9x higher reward and improves training stability when compared to baselines that only use high-fidelity data. Moreover, even when the baselines are given more high-fidelity samples--up to 10x as many interactions with the target environment--MFPG continues to match or outperform them. Finally, we observe that MFPG is capable of training effective policies even when the low-fidelity environment is drastically different from the target environment. MFPG thus not only offers a novel paradigm for efficient sim-to-real transfer but also provides a principled approach to managing the trade-off between policy performance and data collection costs.
Authors: Umberto Borso, Davide Paglieri, Jude Wells, Tim Rockt\"aschel
Abstract: Diffusion models have achieved state-of-the-art performance across multiple domains, with recent advancements extending their applicability to discrete data. However, aligning discrete diffusion models with task-specific preferences remains challenging, particularly in scenarios where explicit reward functions are unavailable. In this work, we introduce Discrete Diffusion DPO (D2-DPO), the first adaptation of Direct Preference Optimization (DPO) to discrete diffusion models formulated as continuous-time Markov chains. Our approach derives a novel loss function that directly fine-tunes the generative process using preference data while preserving fidelity to a reference distribution. We validate D2-DPO on a structured binary sequence generation task, demonstrating that the method effectively aligns model outputs with preferences while maintaining structural validity. Our results highlight that D2-DPO enables controlled fine-tuning without requiring explicit reward models, making it a practical alternative to reinforcement learning-based approaches. Future research will explore extending D2-DPO to more complex generative tasks, including language modeling and protein sequence generation, as well as investigating alternative noise schedules, such as uniform noising, to enhance flexibility across different applications.
Authors: Yifan Wang, Yifei Liu, Yingdong Shi, Changming Li, Anqi Pang, Sibei Yang, Jingyi Yu, Kan Ren
Abstract: Vision Transformer models exhibit immense power yet remain opaque to human understanding, posing challenges and risks for practical applications. While prior research has attempted to demystify these models through input attribution and neuron role analysis, there's been a notable gap in considering layer-level information and the holistic path of information flow across layers. In this paper, we investigate the significance of influential neuron paths within vision Transformers, which is a path of neurons from the model input to output that impacts the model inference most significantly. We first propose a joint influence measure to assess the contribution of a set of neurons to the model outcome. And we further provide a layer-progressive neuron locating approach that efficiently selects the most influential neuron at each layer trying to discover the crucial neuron path from input to output within the target model. Our experiments demonstrate the superiority of our method finding the most influential neuron path along which the information flows, over the existing baseline solutions. Additionally, the neuron paths have illustrated that vision Transformers exhibit some specific inner working mechanism for processing the visual information within the same image category. We further analyze the key effects of these neurons on the image classification task, showcasing that the found neuron paths have already preserved the model capability on downstream tasks, which may also shed some lights on real-world applications like model pruning. The project website including implementation code is available at https://foundation-model-research.github.io/NeuronPath/.
URLs: https://foundation-model-research.github.io/NeuronPath/.
Authors: Alejandro Ortega
Abstract: Recent progress in AI capabilities has heightened concerns that AI systems could pose a threat to national security, for example, by making it easier for malicious actors to perform cyberattacks on critical national infrastructure, or through loss of control of autonomous AI systems. In parallel, federal legislators in the US have proposed nascent 'AI incident regimes' to identify and counter similar threats. In this paper, we consolidate these two trends and present a proposal for a legally mandated post-deployment AI incident regime that aims to counter potential national security threats from AI systems. We start the paper by introducing the concept of 'security-critical' to describe doctors that pose extreme risks to national security, before arguing that 'security-critical' describes civilian nuclear power, aviation, life science dual-use research of concern, and frontier AI development. We then present in detail our AI incident regime proposal, justifying each component of the proposal by demonstrating its similarity to US domestic incident regimes in other 'security-critical' sectors. Finally, we sketch a hypothetical scenario where our proposed AI incident regime deals with an AI cyber incident. Our proposed AI incident regime is split into three phases. The first phase revolves around a novel operationalization of what counts as an 'AI incident' and we suggest that AI providers must create a 'national security case' before deploying a frontier AI system. The second and third phases spell out that AI providers should notify a government agency about incidents, and that the government agency should be involved in amending AI providers' security and safety procedures, in order to counter future threats to national security.
Authors: Iuri Macocco, Nora Graichen, Gemma Boleda, Marco Baroni
Abstract: We study last-layer outlier dimensions, i.e. dimensions that display extreme activations for the majority of inputs. We show that outlier dimensions arise in many different modern language models, and trace their function back to the heuristic of constantly predicting frequent words. We further show how a model can block this heuristic when it is not contextually appropriate, by assigning a counterbalancing weight mass to the remaining dimensions, and we investigate which model parameters boost outlier dimensions and when they arise during training. We conclude that outlier dimensions are a specialized mechanism discovered by many distinct models to implement a useful token prediction heuristic.
Authors: Yijie Zheng, Bangjun Xiao, Lei Shi, Xiaoyang Li, Faming Wu, Tianyu Li, Xuefeng Xiao, Yang Zhang, Yuxuan Wang, Shouda Liu
Abstract: Multimodal large language models (MLLMs), such as GPT-4o, are garnering significant attention. During the exploration of MLLM training, we identified Modality Composition Incoherence, a phenomenon that the proportion of a certain modality varies dramatically across different examples. It exacerbates the challenges of addressing mini-batch imbalances, which lead to uneven GPU utilization between Data Parallel (DP) instances and severely degrades the efficiency and scalability of MLLM training, ultimately affecting training speed and hindering further research on MLLMs. To address these challenges, we introduce OrchMLLM, a comprehensive framework designed to mitigate the inefficiencies in MLLM training caused by Modality Composition Incoherence. First, we propose Batch Post-Balancing Dispatcher, a technique that efficiently eliminates mini-batch imbalances in sequential data. Additionally, we integrate MLLM Global Orchestrator into the training framework to orchestrate multimodal data and tackle the issues arising from Modality Composition Incoherence. We evaluate OrchMLLM across various MLLM sizes, demonstrating its efficiency and scalability. Experimental results reveal that OrchMLLM achieves a Model FLOPs Utilization (MFU) of $41.6\%$ when training an 84B MLLM with three modalities on $2560$ H100 GPUs, outperforming Megatron-LM by up to $3.1\times$ in throughput.
Authors: Rana Muhammad Shahroz Khan, Dongwen Tang, Pingzhi Li, Kai Wang, Tianlong Chen
Abstract: Parameter generation has emerged as a novel paradigm for neural network development, offering an alternative to traditional neural network training by synthesizing high-quality model weights directly. In the context of Low-Rank Adaptation (LoRA) for evolving ($\textit{i.e.}$, constantly updated) large language models (LLMs), this approach promises efficient adaptation without costly retraining. However, existing methods face critical limitations in simultaneously achieving scalability and controllability. In this paper, we introduce $\texttt{ORAL}$, a novel $\textbf{conditional recurrent diffusion}$ framework that addresses these challenges. $\texttt{ORAL}$ incorporates a novel conditioning mechanism that integrates model architecture and textual task specifications, enabling the generation of task-specific LoRA parameters that can seamlessly transfer across evolving foundation models. Our approach successfully scales to billions-of-parameter LLMs and maintains controllability. Through extensive experiments across seven language tasks, four vision tasks, and three multimodal tasks using five pre-trained LLMs, we demonstrate that $\texttt{ORAL}$ generates high-quality LoRA parameters that achieve comparable or superior performance to vanilla trained counterparts.
Authors: Yi Nian, Shenzhe Zhu, Yuehan Qin, Li Li, Ziyi Wang, Chaowei Xiao, Yue Zhao
Abstract: Multimodal large language models (MLLMs) excel in vision-language tasks but also pose significant risks of generating harmful content, particularly through jailbreak attacks. Jailbreak attacks refer to intentional manipulations that bypass safety mechanisms in models, leading to the generation of inappropriate or unsafe content. Detecting such attacks is critical to ensuring the responsible deployment of MLLMs. Existing jailbreak detection methods face three primary challenges: (1) Many rely on model hidden states or gradients, limiting their applicability to white-box models, where the internal workings of the model are accessible; (2) They involve high computational overhead from uncertainty-based analysis, which limits real-time detection, and (3) They require fully labeled harmful datasets, which are often scarce in real-world settings. To address these issues, we introduce a test-time adaptive framework called JAILDAM. Our method leverages a memory-based approach guided by policy-driven unsafe knowledge representations, eliminating the need for explicit exposure to harmful data. By dynamically updating unsafe knowledge during test-time, our framework improves generalization to unseen jailbreak strategies while maintaining efficiency. Experiments on multiple VLM jailbreak benchmarks demonstrate that JAILDAM delivers state-of-the-art performance in harmful content detection, improving both accuracy and speed.
Authors: Kai Ye, Hongyi Zhou, Jin Zhu, Francesco Quinzan, Chengchung Shi
Abstract: Reinforcement learning from human feedback (RLHF) has emerged as a key technique for aligning the output of large language models (LLMs) with human preferences. To learn the reward function, most existing RLHF algorithms use the Bradley-Terry model, which relies on assumptions about human preferences that may not reflect the complexity and variability of real-world judgments. In this paper, we propose a robust algorithm to enhance the performance of existing approaches under such reward model misspecifications. Theoretically, our algorithm reduces the variance of reward and policy estimators, leading to improved regret bounds. Empirical evaluations on LLM benchmark datasets demonstrate that the proposed algorithm consistently outperforms existing methods, with 77-81% of responses being favored over baselines on the Anthropic Helpful and Harmless dataset.
Authors: Marios Kokkodis, Richard Demsyn-Jones, Vijay Raghavan
Abstract: Are traditional classification approaches irrelevant in this era of AI hype? We show that there are multiclass classification problems where predictive models holistically outperform LLM prompt-based frameworks. Given text and images from home-service project descriptions provided by Thumbtack customers, we build embeddings-based softmax models that predict the professional category (e.g., handyman, bathroom remodeling) associated with each problem description. We then compare against prompts that ask state-of-the-art LLM models to solve the same problem. We find that the embeddings approach outperforms the best LLM prompts in terms of accuracy, calibration, latency, and financial cost. In particular, the embeddings approach has 49.5% higher accuracy than the prompting approach, and its superiority is consistent across text-only, image-only, and text-image problem descriptions. Furthermore, it yields well-calibrated probabilities, which we later use as confidence signals to provide contextualized user experience during deployment. On the contrary, prompting scores are overly uninformative. Finally, the embeddings approach is 14 and 81 times faster than prompting in processing images and text respectively, while under realistic deployment assumptions, it can be up to 10 times cheaper. Based on these results, we deployed a variation of the embeddings approach, and through A/B testing we observed performance consistent with our offline analysis. Our study shows that for multiclass classification problems that can leverage proprietary datasets, an embeddings-based approach may yield unequivocally better results. Hence, scientists, practitioners, engineers, and business leaders can use our study to go beyond the hype and consider appropriate predictive models for their classification use cases.
Authors: Yao Tao, Yehui Tang, Yun Wang, Mingjian Zhu, Hailin Hu, Yunhe Wang
Abstract: Despite the recent success of large language models (LLMs), LLMs are particularly challenging in long-sequence inference scenarios due to the quadratic computational complexity of the attention mechanism. Inspired by the interpretability theory of feature attribution in neural network models, we observe that not all tokens have the same contribution. Based on this observation, we propose a novel token pruning framework, namely Saliency-driven Dynamic Token Pruning (SDTP), to gradually and dynamically prune redundant tokens based on the input context. Specifically, a lightweight saliency-driven prediction module is designed to estimate the importance score of each token with its hidden state, which is added to different layers of the LLM to hierarchically prune redundant tokens. Furthermore, a ranking-based optimization strategy is proposed to minimize the ranking divergence of the saliency score and the predicted importance score. Extensive experiments have shown that our framework is generalizable to various models and datasets. By hierarchically pruning 65\% of the input tokens, our method greatly reduces 33\% $\sim$ 47\% FLOPs and achieves speedup up to 1.75$\times$ during inference, while maintaining comparable performance. We further demonstrate that SDTP can be combined with KV cache compression method for further compression.
Authors: Bingyang Wang, Kaer Huang, Bin Li, Yiqiang Yan, Lihe Zhang, Huchuan Lu, You He
Abstract: Open-World Tracking (OWT) aims to track every object of any category, which requires the model to have strong generalization capabilities. Trackers can improve their generalization ability by leveraging Visual Language Models (VLMs). However, challenges arise with the fine-tuning strategies when VLMs are transferred to OWT: full fine-tuning results in excessive parameter and memory costs, while the zero-shot strategy leads to sub-optimal performance. To solve the problem, EffOWT is proposed for efficiently transferring VLMs to OWT. Specifically, we build a small and independent learnable side network outside the VLM backbone. By freezing the backbone and only executing backpropagation on the side network, the model's efficiency requirements can be met. In addition, EffOWT enhances the side network by proposing a hybrid structure of Transformer and CNN to improve the model's performance in the OWT field. Finally, we implement sparse interactions on the MLP, thus reducing parameter updates and memory costs significantly. Thanks to the proposed methods, EffOWT achieves an absolute gain of 5.5% on the tracking metric OWTA for unknown categories, while only updating 1.3% of the parameters compared to full fine-tuning, with a 36.4% memory saving. Other metrics also demonstrate obvious improvement.
Authors: Sanchit Kabra, Akshita Jha, Chandan K. Reddy
Abstract: Recent advances in large-scale generative language models have shown that reasoning capabilities can significantly improve model performance across a variety of tasks. However, the impact of reasoning on a model's ability to mitigate stereotypical responses remains largely underexplored. In this work, we investigate the crucial relationship between a model's reasoning ability and fairness, and ask whether improved reasoning capabilities can mitigate harmful stereotypical responses, especially those arising due to shallow or flawed reasoning. We conduct a comprehensive evaluation of multiple open-source LLMs, and find that larger models with stronger reasoning abilities exhibit substantially lower stereotypical bias on existing fairness benchmarks. Building on this insight, we introduce ReGiFT -- Reasoning Guided Fine-Tuning, a novel approach that extracts structured reasoning traces from advanced reasoning models and infuses them into models that lack such capabilities. We use only general-purpose reasoning and do not require any fairness-specific supervision for bias mitigation. Notably, we see that models fine-tuned using ReGiFT not only improve fairness relative to their non-reasoning counterparts but also outperform advanced reasoning models on fairness benchmarks. We also analyze how variations in the correctness of the reasoning traces and their length influence model fairness and their overall performance. Our findings highlight that enhancing reasoning capabilities is an effective, fairness-agnostic strategy for mitigating stereotypical bias caused by reasoning flaws.
Authors: Thomas Chen, Chun-Kai Kevin Chien, Patricia Mu\~noz Ewald, Andrew G. Moore
Abstract: We prove that overparametrized neural networks are able to generalize with a test error that is independent of the level of overparametrization, and independent of the Vapnik-Chervonenkis (VC) dimension. We prove explicit bounds that only depend on the metric geometry of the test and training sets, on the regularity properties of the activation function, and on the operator norms of the weights and norms of biases. For overparametrized deep ReLU networks with a training sample size bounded by the input space dimension, we explicitly construct zero loss minimizers without use of gradient descent, and prove that the generalization error is independent of the network architecture.
Authors: Shuai Wang, Zhi Tian, Weilin Huang, Limin Wang
Abstract: Diffusion transformers have demonstrated remarkable generation quality, albeit requiring longer training iterations and numerous inference steps. In each denoising step, diffusion transformers encode the noisy inputs to extract the lower-frequency semantic component and then decode the higher frequency with identical modules. This scheme creates an inherent optimization dilemma: encoding low-frequency semantics necessitates reducing high-frequency components, creating tension between semantic encoding and high-frequency decoding. To resolve this challenge, we propose a new \textbf{\color{ddt}D}ecoupled \textbf{\color{ddt}D}iffusion \textbf{\color{ddt}T}ransformer~(\textbf{\color{ddt}DDT}), with a decoupled design of a dedicated condition encoder for semantic extraction alongside a specialized velocity decoder. Our experiments reveal that a more substantial encoder yields performance improvements as model size increases. For ImageNet $256\times256$, Our DDT-XL/2 achieves a new state-of-the-art performance of {1.31 FID}~(nearly $4\times$ faster training convergence compared to previous diffusion transformers). For ImageNet $512\times512$, Our DDT-XL/2 achieves a new state-of-the-art FID of 1.28. Additionally, as a beneficial by-product, our decoupled architecture enhances inference speed by enabling the sharing self-condition between adjacent denoising steps. To minimize performance degradation, we propose a novel statistical dynamic programming approach to identify optimal sharing strategies.
Authors: Bowen Lou, Tian Lu, T. S. Raghu, Yingjie Zhang
Abstract: Artificial Intelligence (AI) is advancing at an unprecedented pace, with clear potential to enhance decision-making and productivity. Yet, the collaborative decision-making process between humans and AI remains underdeveloped, often falling short of its transformative possibilities. This paper explores the evolution of AI agents from passive tools to active collaborators in human-AI teams, emphasizing their ability to learn, adapt, and operate autonomously in complex environments. This paradigm shifts challenges traditional team dynamics, requiring new interaction protocols, delegation strategies, and responsibility distribution frameworks. Drawing on Team Situation Awareness (SA) theory, we identify two critical gaps in current human-AI teaming research: the difficulty of aligning AI agents with human values and objectives, and the underutilization of AI's capabilities as genuine team members. Addressing these gaps, we propose a structured research outlook centered on four key aspects of human-AI teaming: formulation, coordination, maintenance, and training. Our framework highlights the importance of shared mental models, trust-building, conflict resolution, and skill adaptation for effective teaming. Furthermore, we discuss the unique challenges posed by varying team compositions, goals, and complexities. This paper provides a foundational agenda for future research and practical design of sustainable, high-performing human-AI teams.
Authors: Seunghyun Ji, Soowon Lee
Abstract: Masked language modeling is a widely used method for learning language representations, where the model predicts a randomly masked word in each input. However, this approach typically considers only a single correct answer during training, ignoring the variety of plausible alternatives that humans might choose. This issue becomes more pronounced when the input text is short, as the possible word distribution tends to have higher entropy, potentially causing the model to become overconfident in its predictions. To mitigate this, we propose a novel confidence regularizer that adaptively adjusts the regularization strength based on the input length. Experiments on the GLUE and SQuAD benchmarks show that our method improves both accuracy and expected calibration error
Authors: Rijul Magu, Arka Dutta, Sean Kim, Ashiqur R. KhudaBukhsh, Munmun De Choudhury
Abstract: Large Language Models (LLMs) have been shown to demonstrate imbalanced biases against certain groups. However, the study of unprovoked targeted attacks by LLMs towards at-risk populations remains underexplored. Our paper presents three novel contributions: (1) the explicit evaluation of LLM-generated attacks on highly vulnerable mental health groups; (2) a network-based framework to study the propagation of relative biases; and (3) an assessment of the relative degree of stigmatization that emerges from these attacks. Our analysis of a recently released large-scale bias audit dataset reveals that mental health entities occupy central positions within attack narrative networks, as revealed by a significantly higher mean centrality of closeness (p-value = 4.06e-10) and dense clustering (Gini coefficient = 0.7). Drawing from sociological foundations of stigmatization theory, our stigmatization analysis indicates increased labeling components for mental health disorder-related targets relative to initial targets in generation chains. Taken together, these insights shed light on the structural predilections of large language models to heighten harmful discourse and highlight the need for suitable approaches for mitigation.