Nan Zhang

Kunlun Zhu, Yifan Luo, Dingling Xu et al.

Retrieval-Augmented Generation (RAG) is a powerful approach that enables large language models (LLMs) to incorporate external knowledge. However, evaluating the effectiveness of RAG systems in specialized scenarios remains challenging due to the high costs of data construction and the lack of suitable evaluation metrics. This paper introduces RAGEval, a framework designed to assess RAG systems across diverse scenarios by generating high-quality documents, questions, answers, and references through a schema-based pipeline. With a focus on factual accuracy, we propose three novel metrics: Completeness, Hallucination, and Irrelevance to evaluate LLM generated responses rigorously. Experimental results show that RAGEval outperforms zero-shot and one-shot methods in terms of clarity, safety, conformity, and richness of generated samples. Furthermore, the use of LLMs for scoring the proposed metrics demonstrates a high level of consistency with human evaluations. RAGEval establishes a new paradigm for evaluating RAG systems in real-world applications. The code and dataset are released at https://github.com/OpenBMB/RAGEval.

Nan Zhang, Yusen Zhang, Wu Guo et al.

Summaries of medical text shall be faithful by being consistent and factual with source inputs, which is an important but understudied topic for safety and efficiency in healthcare. In this paper, we investigate and improve faithfulness in summarization on a broad range of medical summarization tasks. Our investigation reveals that current summarization models often produce unfaithful outputs for medical input text. We then introduce FaMeSumm, a framework to improve faithfulness by fine-tuning pre-trained language models based on medical knowledge. FaMeSumm performs contrastive learning on designed sets of faithful and unfaithful summaries, and it incorporates medical terms and their contexts to encourage faithful generation of medical terms. We conduct comprehensive experiments on three datasets in two languages: health question and radiology report summarization datasets in English, and a patient-doctor dialogue dataset in Chinese. Results demonstrate that FaMeSumm is flexible and effective by delivering consistent improvements over mainstream language models such as BART, T5, mT5, and PEGASUS, yielding state-of-the-art performances on metrics for faithfulness and general quality. Human evaluation by doctors also shows that FaMeSumm generates more faithful outputs. Our code is available at https://github.com/psunlpgroup/FaMeSumm .

Zixian Su, Hongkai Zhang, Fan Gao et al.

Multimodal Large Language Models (MLLMs) have shown strong performance on public medical benchmarks, yet existing evaluations often remain weak proxies for clinical use, relying on isolated inputs and simplified recognition-style tasks. We introduce CardioLens, a leakage-resistant evaluation testbed for multi-sequence Cardiovascular Magnetic Resonance (CMR), constructed from private hospital archives through a rigorous report-to-QA construction and verification pipeline. CardioLens contains 473,896 slices and 13,494 verified QA pairs across 4D Cine, LGE, perfusion, and T2-weighted imaging, and evaluates three stages of CMR interpretation: image understanding, report generation, and disease diagnosis. Across 24 state-of-the-art MLLMs, CardioLens reveals a substantial clinical reality gap: models perform poorly overall, with performance degrading along the real CMR workflow. Confusion analysis further shows a category-collapse failure mode, where models default to frequent abnormal categories rather than distinguishing clinically distinct findings. To rule out MLLM-compatible input construction as the primary cause, we compare random, clinically motivated, and data-driven slice selection protocols under different slice budgets; performance changes only marginally, typically by about 1%. Explicit reasoning prompts also fail to rescue performance, often making models more conservative rather than improving visual evidence use. These results show that current MLLMs remain far from reliable CMR interpretation, where clinical decisions require integrating distributed evidence across sequences, views, and temporal phases. CardioLens provides a clinically grounded testbed for developing next-generation MLLMs toward real-world clinical deployment.

Yusen Zhang, Nan Zhang, Yixin Liu et al.

People from different social and demographic groups express diverse perspectives and conflicting opinions on a broad set of topics such as product reviews, healthcare, law, and politics. A fair summary should provide a comprehensive coverage of diverse perspectives without underrepresenting certain groups. However, current work in summarization metrics and Large Language Models (LLMs) evaluation has not explored fair abstractive summarization. In this paper, we systematically investigate fair abstractive summarization for user-generated data. We first formally define fairness in abstractive summarization as not underrepresenting perspectives of any groups of people, and we propose four reference-free automatic metrics by measuring the differences between target and source perspectives. We evaluate nine LLMs, including three GPT models, four LLaMA models, PaLM 2, and Claude, on six datasets collected from social media, online reviews, and recorded transcripts. Experiments show that both the model-generated and the human-written reference summaries suffer from low fairness. We conduct a comprehensive analysis of the common factors influencing fairness and propose three simple but effective methods to alleviate unfair summarization. Our dataset and code are available at https://github.com/psunlpgroup/FairSumm.

Liang Wang, Kai Lu, Nan Zhang et al.

This paper proposes Shoggoth, an efficient edge-cloud collaborative architecture, for boosting inference performance on real-time video of changing scenes. Shoggoth uses online knowledge distillation to improve the accuracy of models suffering from data drift and offloads the labeling process to the cloud, alleviating constrained resources of edge devices. At the edge, we design adaptive training using small batches to adapt models under limited computing power, and adaptive sampling of training frames for robustness and reducing bandwidth. The evaluations on the realistic dataset show 15%-20% model accuracy improvement compared to the edge-only strategy and fewer network costs than the cloud-only strategy.

Hao Zhang, Nan Zhang, Ruixin Zhang et al. · tencent-ai

In recent years, molecular graph representation learning (GRL) has drawn much more attention in molecular property prediction (MPP) problems. The existing graph methods have demonstrated that 3D geometric information is significant for better performance in MPP. However, accurate 3D structures are often costly and time-consuming to obtain, limiting the large-scale application of GRL. It is an intuitive solution to train with 3D to 2D knowledge distillation and predict with only 2D inputs. But some challenging problems remain open for 3D to 2D distillation. One is that the 3D view is quite distinct from the 2D view, and the other is that the gradient magnitudes of atoms in distillation are discrepant and unstable due to the variable molecular size. To address these challenging problems, we exclusively propose a distillation framework that contains global molecular distillation and local atom distillation. We also provide a theoretical insight to justify how to coordinate atom and molecular information, which tackles the drawback of variable molecular size for atom information distillation. Experimental results on two popular molecular datasets demonstrate that our proposed model achieves superior performance over other methods. Specifically, on the largest MPP dataset PCQM4Mv2 served as an "ImageNet Large Scale Visual Recognition Challenge" in the field of graph ML, the proposed method achieved a 6.9% improvement compared with the best works. And we obtained fourth place with the MAE of 0.0734 on the test-challenge set for OGB-LSC 2022 Graph Regression Task. We will release the code soon.

Liang Wang, Nan Zhang, Xiaoyang Qu et al.

Real-time video analytics on edge devices for changing scenes remains a difficult task. As edge devices are usually resource-constrained, edge deep neural networks (DNNs) have fewer weights and shallower architectures than general DNNs. As a result, they only perform well in limited scenarios and are sensitive to data drift. In this paper, we introduce EdgeMA, a practical and efficient video analytics system designed to adapt models to shifts in real-world video streams over time, addressing the data drift problem. EdgeMA extracts the gray level co-occurrence matrix based statistical texture feature and uses the Random Forest classifier to detect the domain shift. Moreover, we have incorporated a method of model adaptation based on importance weighting, specifically designed to update models to cope with the label distribution shift. Through rigorous evaluation of EdgeMA on a real-world dataset, our results illustrate that EdgeMA significantly improves inference accuracy.

Jifan Yu, Zheyuan Zhang, Daniel Zhang-li et al.

Since the first instances of online education, where courses were uploaded to accessible and shared online platforms, this form of scaling the dissemination of human knowledge to reach a broader audience has sparked extensive discussion and widespread adoption. Recognizing that personalized learning still holds significant potential for improvement, new AI technologies have been continuously integrated into this learning format, resulting in a variety of educational AI applications such as educational recommendation and intelligent tutoring. The emergence of intelligence in large language models (LLMs) has allowed for these educational enhancements to be built upon a unified foundational model, enabling deeper integration. In this context, we propose MAIC (Massive AI-empowered Course), a new form of online education that leverages LLM-driven multi-agent systems to construct an AI-augmented classroom, balancing scalability with adaptivity. Beyond exploring the conceptual framework and technical innovations, we conduct preliminary experiments at Tsinghua University, one of China's leading universities. Drawing from over 100,000 learning records of more than 500 students, we obtain a series of valuable observations and initial analyses. This project will continue to evolve, ultimately aiming to establish a comprehensive open platform that supports and unifies research, technology, and applications in exploring the possibilities of online education in the era of large model AI. We envision this platform as a collaborative hub, bringing together educators, researchers, and innovators to collectively explore the future of AI-driven online education.

Junlin Hou, Jilan Xu, Nan Zhang et al.

This paper presents our solution for the 2nd COVID-19 Severity Detection Competition. This task aims to distinguish the Mild, Moderate, Severe, and Critical grades in COVID-19 chest CT images. In our approach, we devise a novel infection-aware 3D Contrastive Mixup Classification network for severity grading. Specifcally, we train two segmentation networks to first extract the lung region and then the inner lesion region. The lesion segmentation mask serves as complementary information for the original CT slices. To relieve the issue of imbalanced data distribution, we further improve the advanced Contrastive Mixup Classification network by weighted cross-entropy loss. On the COVID-19 severity detection leaderboard, our approach won the first place with a Macro F1 Score of 51.76%. It significantly outperforms the baseline method by over 11.46%.

Junlin Hou, Jilan Xu, Nan Zhang et al.

This paper presents our solution for the 2nd COVID-19 Competition, occurring in the framework of the AIMIA Workshop at the European Conference on Computer Vision (ECCV 2022). In our approach, we employ the winning solution last year which uses a strong 3D Contrastive Mixup Classifcation network (CMC v1) as the baseline method, composed of contrastive representation learning and mixup classification. In this paper, we propose CMC v2 by introducing natural video priors to COVID-19 diagnosis. Specifcally, we adapt a pre-trained (on video dataset) video transformer backbone to COVID-19 detection. Moreover, advanced training strategies, including hybrid mixup and cutmix, slicelevel augmentation, and small resolution training are also utilized to boost the robustness and the generalization ability of the model. Among 14 participating teams, CMC v2 ranked 1st in the 2nd COVID-19 Competition with an average Macro F1 Score of 89.11%.

Nan Zhang, Rami Bahsoon, Nikos Tziritas et al.

A digital twin contains up-to-date data-driven models of the physical world being studied and can use simulation to optimise the physical world. However, the analysis made by the digital twin is valid and reliable only when the model is equivalent to the physical world. Maintaining such an equivalent model is challenging, especially when the physical systems being modelled are intelligent and autonomous. The paper focuses in particular on digital twin models of intelligent systems where the systems are knowledge-aware but with limited capability. The digital twin improves the acting of the physical system at a meta-level by accumulating more knowledge in the simulated environment. The modelling of such an intelligent physical system requires replicating the knowledge-awareness capability in the virtual space. Novel equivalence maintaining techniques are needed, especially in synchronising the knowledge between the model and the physical system. This paper proposes the notion of knowledge equivalence and an equivalence maintaining approach by knowledge comparison and updates. A quantitative analysis of the proposed approach confirms that compared to state equivalence, knowledge equivalence maintenance can tolerate deviation thus reducing unnecessary updates and achieve more Pareto efficient solutions for the trade-off between update overhead and simulation reliability.

Nan Zhang, Rami Bahsoon, Nikos Tziritas et al.

Digital Twins (DT) are essentially dynamic data-driven models that serve as real-time symbiotic "virtual replicas" of real-world systems. DT can leverage fundamentals of Dynamic Data-Driven Applications Systems (DDDAS) bidirectional symbiotic sensing feedback loops for its continuous updates. Sensing loops can consequently steer measurement, analysis and reconfiguration aimed at more accurate modelling and analysis in DT. The reconfiguration decisions can be autonomous or interactive, keeping human-in-the-loop. The trustworthiness of these decisions can be hindered by inadequate explainability of the rationale, and utility gained in implementing the decision for the given situation among alternatives. Additionally, different decision-making algorithms and models have varying complexity, quality and can result in different utility gained for the model. The inadequacy of explainability can limit the extent to which humans can evaluate the decisions, often leading to updates which are unfit for the given situation, erroneous, compromising the overall accuracy of the model. The novel contribution of this paper is an approach to harnessing explainability in human-in-the-loop DDDAS and DT systems, leveraging bidirectional symbiotic sensing feedback. The approach utilises interpretable machine learning and goal modelling to explainability, and considers trade-off analysis of utility gained. We use examples from smart warehousing to demonstrate the approach.

Muyu Pan, Shu Zhao, Nan Zhang et al.

This paper investigates large language model (LLM) abstention learning, specifically using ternary reward, which incentivize truthfulness in large language models. This paper extends that idea by moving from a ternary reward to a Trajectory-Informed advantage reweighting, dynamically re-weights the abstention reward during Group Relative Policy Optimization (GRPO) training. The objective of this work focuses on abstention learning instead of improving truthfulness, serving as an exploration into hallucination reduction. The novelty of this paper lies in methodological innovation, advantage re-weighting, and benchmark selection. Leveraging GRPO's multiple trajectories as a natural abstention signal, this method uses a reward signal to explore knowledge boundaries and encourage consistency. By demonstrating that trajectories can be used as a confidence indicator of the policy relative to the query, they are then used to dynamically calculate the abstention advantage. AbstentionBench is used as the evaluation benchmark, as this work aims to contribute to the field of abstention learning. All datasets on the benchmark were tested against this method and various baselines. Empirical results demonstrate that TIAR achieves state-of-the-art abstention F1 scores across five of six evaluation categories, outperforming the static ternary baseline on 17 of 31 benchmark datasets while fully preserving baseline accuracy.

Nan Zhang, Jianzong Wang, Zhenhou Hong et al.

Speaker verification (SV) aims to determine whether the speaker's identity of a test utterance is the same as the reference speech. In the past few years, extracting speaker embeddings using deep neural networks for SV systems has gone mainstream. Recently, different attention mechanisms and Transformer networks have been explored widely in SV fields. However, utilizing the original Transformer in SV directly may have frame-level information waste on output features, which could lead to restrictions on capacity and discrimination of speaker embeddings. Therefore, we propose an approach to derive utterance-level speaker embeddings via a Transformer architecture that uses a novel loss function named diffluence loss to integrate the feature information of different Transformer layers. Therein, the diffluence loss aims to aggregate frame-level features into an utterance-level representation, and it could be integrated into the Transformer expediently. Besides, we also introduce a learnable mel-fbank energy feature extractor named time-domain feature extractor that computes the mel-fbank features more precisely and efficiently than the standard mel-fbank extractor. Combining Diffluence loss and Time-domain feature extractor, we propose a novel Transformer-based time-domain SV model (DT-SV) with faster training speed and higher accuracy. Experiments indicate that our proposed model can achieve better performance in comparison with other models.

Liang Shi, Zhengju Tang, Nan Zhang et al.

With the development of the Large Language Models (LLMs), a large range of LLM-based Text-to-SQL(Text2SQL) methods have emerged. This survey provides a comprehensive review of LLM-based Text2SQL studies. We first enumerate classic benchmarks and evaluation metrics. For the two mainstream methods, prompt engineering and finetuning, we introduce a comprehensive taxonomy and offer practical insights into each subcategory. We present an overall analysis of the above methods and various models evaluated on well-known datasets and extract some characteristics. Finally, we discuss the challenges and future directions in this field.

Ryo Kamoi, Sarkar Snigdha Sarathi Das, Renze Lou et al.

With Large Language Models (LLMs) being widely used across various tasks, detecting errors in their responses is increasingly crucial. However, little research has been conducted on error detection of LLM responses. Collecting error annotations on LLM responses is challenging due to the subjective nature of many NLP tasks, and thus previous research focuses on tasks of little practical value (e.g., word sorting) or limited error types (e.g., faithfulness in summarization). This work introduces ReaLMistake, the first error detection benchmark consisting of objective, realistic, and diverse errors made by LLMs. ReaLMistake contains three challenging and meaningful tasks that introduce objectively assessable errors in four categories (reasoning correctness, instruction-following, context-faithfulness, and parameterized knowledge), eliciting naturally observed and diverse errors in responses of GPT-4 and Llama 2 70B annotated by experts. We use ReaLMistake to evaluate error detectors based on 12 LLMs. Our findings show: 1) Top LLMs like GPT-4 and Claude 3 detect errors made by LLMs at very low recall, and all LLM-based error detectors perform much worse than humans. 2) Explanations by LLM-based error detectors lack reliability. 3) LLMs-based error detection is sensitive to small changes in prompts but remains challenging to improve. 4) Popular approaches to improving LLMs, including self-consistency and majority vote, do not improve the error detection performance. Our benchmark and code are provided at https://github.com/psunlpgroup/ReaLMistake.

Shuang Li, Ziyuan Pu, Nan Zhang et al.

Traffic crashes profoundly impede traffic efficiency and pose economic challenges. Accurate prediction of post-crash traffic status provides essential information for evaluating traffic perturbations and developing effective solutions. Previous studies have established a series of deep learning models to predict post-crash traffic conditions, however, these correlation-based methods cannot accommodate the biases caused by time-varying confounders and the heterogeneous effects of crashes. The post-crash traffic prediction model needs to estimate the counterfactual traffic speed response to hypothetical crashes under various conditions, which demonstrates the necessity of understanding the causal relationship between traffic factors. Therefore, this paper presents the Marginal Structural Causal Transformer (MSCT), a novel deep learning model designed for counterfactual post-crash traffic prediction. To address the issue of time-varying confounding bias, MSCT incorporates a structure inspired by Marginal Structural Models and introduces a balanced loss function to facilitate learning of invariant causal features. The proposed model is treatment-aware, with a specific focus on comprehending and predicting traffic speed under hypothetical crash intervention strategies. In the absence of ground-truth data, a synthetic data generation procedure is proposed to emulate the causal mechanism between traffic speed, crashes, and covariates. The model is validated using both synthetic and real-world data, demonstrating that MSCT outperforms state-of-the-art models in multi-step-ahead prediction performance. This study also systematically analyzes the impact of time-varying confounding bias and dataset distribution on model performance, contributing valuable insights into counterfactual prediction for intelligent transportation systems.

Nan Zhang, Yanchi Liu, Xujiang Zhao et al.

Large Language Models (LLMs) have exhibited remarkable proficiency across a wide array of NLP tasks. However, the escalation in model size also engenders substantial deployment costs. While few efforts have explored model pruning techniques to reduce the size of LLMs, they mainly center on general or task-specific weights. This leads to suboptimal performance due to lacking specificity on the target domain or generality on different tasks when applied to domain-specific challenges. This work introduces an innovative unstructured dual-pruning methodology, D-Pruner, for domain-specific compression on LLM. It extracts a compressed, domain-specific, and task-agnostic LLM by identifying LLM weights that are pivotal for general capabilities, like linguistic capability and multi-task solving, and domain-specific knowledge. More specifically, we first assess general weight importance by quantifying the error incurred upon their removal with the help of an open-domain calibration dataset. Then, we utilize this general weight importance to refine the training loss, so that it preserves generality when fitting into a specific domain. Moreover, by efficiently approximating weight importance with the refined training loss on a domain-specific calibration dataset, we obtain a pruned model emphasizing generality and specificity. Our comprehensive experiments across various tasks in healthcare and legal domains show the effectiveness of D-Pruner in domain-specific compression. Our code is available at https://github.com/psunlpgroup/D-Pruner.

Nan Zhang, Prafulla Kumar Choubey, Alexander Fabbri et al.

Indexing is an important step towards strong performance in retrieval-augmented generation (RAG) systems. However, existing methods organize data based on either semantic similarity (similarity) or related information (relatedness), but do not cover both perspectives comprehensively. Our analysis reveals that modeling only one perspective results in insufficient knowledge synthesis, leading to suboptimal performance on complex tasks requiring multihop reasoning. In this paper, we propose SiReRAG, a novel RAG indexing approach that explicitly considers both similar and related information. On the similarity side, we follow existing work and explore some variances to construct a similarity tree based on recursive summarization. On the relatedness side, SiReRAG extracts propositions and entities from texts, groups propositions via shared entities, and generates recursive summaries to construct a relatedness tree. We index and flatten both similarity and relatedness trees into a unified retrieval pool. Our experiments demonstrate that SiReRAG consistently outperforms state-of-the-art indexing methods on three multihop datasets (MuSiQue, 2WikiMultiHopQA, and HotpotQA), with an average 1.9% improvement in F1 scores. As a reasonably efficient solution, SiReRAG enhances existing reranking methods significantly, with up to 7.8% improvement in average F1 scores. Our code is available at https://github.com/SalesforceAIResearch/SiReRAG .

Yizhu Wen, Shuhao Zhang, Nan Zhang et al.

Retrieval-augmented text-to-music (TTM) systems augment underspecified user prompts using captions retrieved from a music caption dataset. This design introduces an integrity dependency on the music knowledge database. We show that an attacker can poison the database by injecting a small number of crafted music captions, causing the system to retrieve malicious captions that bias prompt augmentation and steer generation away from the user's intended function, without modifying the user prompt, retriever, or generator. To achieve the music caption poisoning attack, we propose a dual-layer caption poisoning strategy that preserves high-level retrieval anchors while injecting low-level acoustic descriptors to steer prompt augmentation and downstream music generation toward an attacker-chosen target intent. In a MusicCaps knowledge database, CLAP retriever, and MusicGen pipeline, poisoned generations move substantially closer to the attacker's target, while remaining comparably aligned with the original user query. These results expose a practical integrity risk for retrieval-augmented creative AI systems. Our demo can be found at: https://yizhu-wen.github.io/Mental-Damage/

Haocheng Lu, Nan Zhang, Wei Tao et al.

Streaming video question answering (Streaming Video QA) poses distinct challenges for multimodal large language models (MLLMs), as video frames arrive sequentially and user queries can be issued at arbitrary time points. Existing solutions relying on fixed-size memory or naive compression often suffer from context loss or memory overflow, limiting their effectiveness in long-form, real-time scenarios. We present Vista, a novel framework for scene-aware streaming video QA that enables efficient and scalable reasoning over continuous video streams. The innovation of Vista can be summarized in three aspects: (1) scene-aware segmentation, where Vista dynamically clusters incoming frames into temporally and visually coherent scene units; (2) scene-aware compression, where each scene is compressed into a compact token representation and stored in GPU memory for efficient index-based retrieval, while full-resolution frames are offloaded to CPU memory; and (3) scene-aware recall, where relevant scenes are selectively recalled and reintegrated into the model input upon receiving a query, enabling both efficiency and completeness. Vista is model-agnostic and integrates seamlessly with a variety of vision-language backbones, enabling long-context reasoning without compromising latency or memory efficiency. Extensive experiments on StreamingBench demonstrate that Vista achieves state-of-the-art performance, establishing a strong baseline for real-world streaming video understanding.

Haojun Yu, Youcheng Li, Nan Zhang et al.

Data-driven deep learning models have shown great capabilities to assist radiologists in breast ultrasound (US) diagnoses. However, their effectiveness is limited by the long-tail distribution of training data, which leads to inaccuracies in rare cases. In this study, we address a long-standing challenge of improving the diagnostic model performance on rare cases using long-tailed data. Specifically, we introduce a pipeline, TAILOR, that builds a knowledge-driven generative model to produce tailored synthetic data. The generative model, using 3,749 lesions as source data, can generate millions of breast-US images, especially for error-prone rare cases. The generated data can be further used to build a diagnostic model for accurate and interpretable diagnoses. In the prospective external evaluation, our diagnostic model outperforms the average performance of nine radiologists by 33.5% in specificity with the same sensitivity, improving their performance by providing predictions with an interpretable decision-making process. Moreover, on ductal carcinoma in situ (DCIS), our diagnostic model outperforms all radiologists by a large margin, with only 34 DCIS lesions in the source data. We believe that TAILOR can potentially be extended to various diseases and imaging modalities.

Xiaohan Wang, Nan Zhang, Sulene Han et al.

The pharmaceutical industry is facing challenges with quality management such as high costs of compliance, slow responses and disjointed knowledge. This paper presents GMPilot, a domain-specific AI agent that is designed to support FDA cGMP compliance. GMPilot is based on a curated knowledge base of regulations and historical inspection observations and uses Retrieval-Augmented Generation (RAG) and Reasoning-Acting (ReAct) frameworks to provide real-time and traceable decision support to the quality professionals. In a simulated inspection scenario, GMPilot shows how it can improve the responsiveness and professionalism of quality professionals by providing structured knowledge retrieval and verifiable regulatory and case-based support. Although GMPilot lacks in the aspect of regulatory scope and model interpretability, it is a viable avenue of improving quality management decision-making in the pharmaceutical sector using intelligent approaches and an example of specialized application of AI in highly regulated sectors.

Anmin Wang, Nan Zhang, Wei Tao et al.

Vision-Language Models (VLMs) face significant computational challenges in video processing due to massive data redundancy, which creates prohibitively long token sequences. To address this, we introduce Triage, a training-free, plug-and-play framework that reframes video reasoning as a resource allocation problem via hierarchical visual budgeting. Its first stage, Frame-Level Budgeting, identifies keyframes by evaluating their visual dynamics and relevance, generating a strategic prior based on their importance scores. Guided by this prior, the second stage, Token-Level Budgeting, allocates tokens in two phases: it first secures high-relevance Core Tokens, followed by diverse Context Tokens selected with an efficient batched Maximal Marginal Relevance (MMR) algorithm. Extensive experiments demonstrate that Triage improves inference speed and reduces memory footprint, while maintaining or surpassing the performance of baselines and other methods on various video reasoning benchmarks.

Ngoc-Hieu Nguyen, Parshin Shojaee, Phuc Minh Nguyen et al.

Recent progress in large language models has led to the emergence of reasoning models, which have shown strong performance on complex tasks through specialized fine-tuning procedures. While these methods reliably improve pass@1 accuracy, prior works have observed that they show a coverage shrinkage behavior, where pass@k degrades relative to the base model. In this paper, we investigate the reasoning shrinkage arise under SFT-based post-training. We hypothesize that this behavior is driven by properties of the fine-tuning data, specifically related to decision points or "forks in the road" scenarios where model faces indecipherable patterns with multiple valid reasoning paths. To test this hypothesis, we design controlled case studies that simulate such decision-point settings, spanning indecipherable nodes in graph branching, and reasoning modes. By tracking post-training dynamics in these settings, we find that the shrinkage phenomenon is tightly correlated with the prevalence of decision-point scenarios in the training data. We also demonstrate that this shrinkage behavior can be partially mitigated through targeted data synthesis design of decision-points, and a more systematic diversity-encouraging decoding mechanism. Our findings identify data-centric factors as a key driver of shrinkage in reasoning models and highlight diversity-aware designs as an effective lever for controlling it.

Nan Zhang, Zishuo Wang, Shuyu Huang et al.

Decentralised online learning enables runtime adaptation in cyber-physical multi-agent systems, but when operating conditions change, learned policies often require substantial trial-and-error interaction before recovering performance. To address this, we propose TwinLoop, a simulation-in-the-loop digital twin framework for online multi-agent reinforcement learning. When a context shift occurs, the digital twin is triggered to reconstruct the current system state, initialise from the latest agent policies, and perform accelerated policy improvement with simulation what-if analysis before synchronising updated parameters back to the agents in the physical system. We evaluate TwinLoop in a vehicular edge computing task-offloading scenario with changing workload and infrastructure conditions. The results suggest that digital twins can improve post-shift adaptation efficiency and reduce reliance on costly online trial-and-error.

Ryo Kamoi, Yusen Zhang, Nan Zhang et al.

Process Reward Models (PRMs), which provide step-level feedback on reasoning traces generated by Large Language Models (LLMs), are receiving increasing attention. However, two key research gaps remain: creating PRM training data requires costly human annotation to label accurate step-level errors, and existing PRMs are limited to math reasoning domains. In response to these gaps, this paper aims to enable automatic synthesis of accurate PRM training data and the generalization of PRMs to diverse reasoning tasks beyond math reasoning. We propose FoVer, an approach to synthesize PRM training data with accurate step-level error labels automatically annotated by formal verification tools, such as Z3 and Isabelle. To show the practical effectiveness of FoVer, we synthesize a training dataset by annotating step-level error labels on LLM responses to formal logic and theorem proving tasks, without relying on human annotation. While FoVer creates training data with symbolic tasks compatible with formal verification, our experiments show that PRMs trained on our dataset exhibit cross-task generalization, enabling a single PRM to effectively perform verification across diverse reasoning tasks. Specifically, LLM-based PRMs trained with FoVer significantly outperform PRMs based on the original LLMs and achieve competitive or superior results compared to state-of-the-art PRMs, as measured by step-level verification on ProcessBench and Best-of-K performance across 12 reasoning benchmarks, including MATH, AIME, ANLI, MMLU, and BBH. The dataset and code are in the supplementary material and will be made public. The datasets, models, and code are provided at https://github.com/psunlpgroup/FoVer.

Taiping Qu, Hongkai Zhang, Lantian Zhang et al.

Cardiac magnetic resonance (CMR) is a cornerstone for diagnosing cardiovascular disease. However, it remains underutilized due to complex, time-consuming interpretation across multi-sequences, phases, quantitative measures that heavily reliant on specialized expertise. Here, we present BAAI Cardiac Agent, a multimodal intelligent system designed for end-to-end CMR interpretation. The agent integrates specialized cardiac expert models to perform automated segmentation of cardiac structures, functional quantification, tissue characterization and disease diagnosis, and generates structured clinical reports within a unified workflow. Evaluated on CMR datasets from two hospitals (2413 patients) spanning 7-types of major cardiovascular diseases, the agent achieved an area under the receiver-operating-characteristic curve exceeding 0.93 internally and 0.81 externally. In the task of estimating left ventricular function indices, the results generated by this system for core parameters such as ejection fraction, stroke volume, and left ventricular mass are highly consistent with clinical reports, with Pearson correlation coefficients all exceeding 0.90. The agent outperformed state-of-the-art models in segmentation and diagnostic tasks, and generated clinical reports showing high concordance with expert radiologists (six readers across three experience levels). By dynamically orchestrating expert models for coordinated multimodal analysis, this agent framework enables accurate, efficient CMR interpretation and highlights its potentials for complex clinical imaging workflows. Code is available at https://github.com/plantain-herb/Cardiac-Agent.

Nan Zhang, Connor Heaton, Sean Timothy Okonsky et al.

Optical Character Recognition (OCR) is an established task with the objective of identifying the text present in an image. While many off-the-shelf OCR models exist, they are often trained for either scientific (e.g., formulae) or generic printed English text. Extracting text from chemistry publications requires an OCR model that is capable in both realms. Nougat, a recent tool, exhibits strong ability to parse academic documents, but is unable to parse tables in PubMed articles, which comprises a significant part of the academic community and is the focus of this work. To mitigate this gap, we present the Printed English and Chemical Equations (PEaCE) dataset, containing both synthetic and real-world records, and evaluate the efficacy of transformer-based OCR models when trained on this resource. Given that real-world records contain artifacts not present in synthetic records, we propose transformations that mimic such qualities. We perform a suite of experiments to explore the impact of patch size, multi-domain training, and our proposed transformations, ultimately finding that models with a small patch size trained on multiple domains using the proposed transformations yield the best performance. Our dataset and code is available at https://github.com/ZN1010/PEaCE.

Nan Zhang, Xidan Zhang, Jianing Wei et al.

This report describes the winning solution to the WeatherProof Dataset Challenge (CVPR 2024 UG2+ Track 3). Details regarding the challenge are available at https://cvpr2024ug2challenge.github.io/track3.html. We propose an enhanced semantic segmentation pipeline for this challenge. Firstly, we improve semantic segmentation models, using backbone pretrained with Depth Anything to improve UperNet model and SETRMLA model, and adding language guidance based on both weather and category information to InternImage model. Secondly, we introduce a new dataset WeatherProofExtra with wider viewing angle and employ data augmentation methods, including adverse weather and super-resolution. Finally, effective training strategies and ensemble method are applied to improve final performance further. Our solution is ranked 1st on the final leaderboard. Code will be available at https://github.com/KaneiGi/WeatherProofChallenge.

Guangji Bai, Zheng Chai, Chen Ling et al.

The burgeoning field of Large Language Models (LLMs), exemplified by sophisticated models like OpenAI's ChatGPT, represents a significant advancement in artificial intelligence. These models, however, bring forth substantial challenges in the high consumption of computational, memory, energy, and financial resources, especially in environments with limited resource capabilities. This survey aims to systematically address these challenges by reviewing a broad spectrum of techniques designed to enhance the resource efficiency of LLMs. We categorize methods based on their optimization focus: computational, memory, energy, financial, and network resources and their applicability across various stages of an LLM's lifecycle, including architecture design, pretraining, finetuning, and system design. Additionally, the survey introduces a nuanced categorization of resource efficiency techniques by their specific resource types, which uncovers the intricate relationships and mappings between various resources and corresponding optimization techniques. A standardized set of evaluation metrics and datasets is also presented to facilitate consistent and fair comparisons across different models and techniques. By offering a comprehensive overview of the current sota and identifying open research avenues, this survey serves as a foundational reference for researchers and practitioners, aiding them in developing more sustainable and efficient LLMs in a rapidly evolving landscape.

Naifu Zhang, Wei Tao, Xi Xiao et al.

In recent years, Vision-Language-Action (VLA) models in embodied intelligence have developed rapidly. However, existing adversarial attack methods require costly end-to-end training and often generate noticeable perturbation patches. To address these limitations, we propose ADVLA, a framework that directly applies adversarial perturbations on features projected from the visual encoder into the textual feature space. ADVLA efficiently disrupts downstream action predictions under low-amplitude constraints, and attention guidance allows the perturbations to be both focused and sparse. We introduce three strategies that enhance sensitivity, enforce sparsity, and concentrate perturbations. Experiments demonstrate that under an $L_{\infty}=4/255$ constraint, ADVLA combined with Top-K masking modifies less than 10% of the patches while achieving an attack success rate of nearly 100%. The perturbations are concentrated on critical regions, remain almost imperceptible in the overall image, and a single-step iteration takes only about 0.06 seconds, significantly outperforming conventional patch-based attacks. In summary, ADVLA effectively weakens downstream action predictions of VLA models under low-amplitude and locally sparse conditions, avoiding the high training costs and conspicuous perturbations of traditional patch attacks, and demonstrates unique effectiveness and practical value for attacking VLA feature spaces.

Xianwei Mao, Kai Ye, Sheng Zhou et al.

Knowledge-based Visual Question Answering (KB-VQA) requires models to answer questions by integrating visual information with external knowledge. However, retrieved knowledge is often noisy, partially irrelevant, or misaligned with the visual content, while internal model knowledge is difficult to control and interpret. Naive aggregation of these sources limits reasoning effectiveness and reduces answer accuracy. To address this, we propose MaS-VQA, a selection-driven framework that tightly couples explicit knowledge filtering with implicit knowledge reasoning. MaS-VQA first retrieves candidate passages and applies a Mask-and-Select mechanism to jointly prune irrelevant image regions and weakly relevant knowledge fragments, producing compact, high-signal multimodal knowledge . This filtered knowledge then guides the activation of internal knowledge in a constrained semantic space, enabling complementary co-modeling of explicit and implicit knowledge for robust answer prediction. Experiments on Encyclopedic-VQA and InfoSeek demonstrate consistent performance gains across multiple MLLM backbones, and ablations verify that the selection mechanism effectively reduces noise and enhances knowledge utilization.

Alina Chertock, Alexander Kurganov, Lorenzo Micalizzi et al.

We propose a new second-order asymptotic-preserving (AP) dual formulation finite-volume (DF-FV) method for the thermal rotating shallow water (TRSW) equations. The TRSW system models geophysical flows characterized by horizontal temperature/density variations, exhibiting multi-scale dynamics due to the coexistence of fast rotational waves and slower advective processes. To efficiently address challenges associated with the multiscale nature of the TRSW system, we follow the DF-FV framework and develop a DF-FV method, in which both the conservative and nonconservative (primitive) forms of the equations are simultaneously solved, allowing the method to exploit the complementary strengths of each representation across different flow regimes. The primitive formulation is better suited for preserving the correct asymptotic behavior in nearly thermal quasi-geostrophic (TQG) regimes characterized by a low Rossby number, while the conservative formulation is essential for robust shock capturing in high-Rossby-number regimes, in which nonconservative discretizations may fail to converge to physically relevant weak solutions.

Qinyu Zhang, Liang Xu, Jianhao Huang et al.

Driven by the vision of ubiquitous connectivity and wireless intelligence, the evolution of ultra-dense constellation-based satellite-integrated Internet is underway, now taking preliminary shape. Nevertheless, the entrenched institutional silos and limited, nonrenewable heterogeneous network resources leave current satellite systems struggling to accommodate the escalating demands of next-generation intelligent applications. In this context, the distributed satellite information networks (DSIN), exemplified by the cohesive clustered satellites system, have emerged as an innovative architecture, bridging information gaps across diverse satellite systems, such as communication, navigation, and remote sensing, and establishing a unified, open information network paradigm to support resilient space information services. This survey first provides a profound discussion about innovative network architectures of DSIN, encompassing distributed regenerative satellite network architecture, distributed satellite computing network architecture, and reconfigurable satellite formation flying, to enable flexible and scalable communication, computing and control. The DSIN faces challenges from network heterogeneity, unpredictable channel dynamics, sparse resources, and decentralized collaboration frameworks. To address these issues, a series of enabling technologies is identified, including channel modeling and estimation, cloud-native distributed MIMO cooperation, grant-free massive access, network routing, and the proper combination of all these diversity techniques. Furthermore, to heighten the overall resource efficiency, the cross-layer optimization techniques are further developed to meet upper-layer deterministic, adaptive and secure information services requirements. In addition, emerging research directions and new opportunities are highlighted on the way to achieving the DSIN vision.

Nan Zhang, Christian Vergara-Marcillo, Georgios Diamantopoulos et al.

Dynamic data-driven Digital Twins (DDTs) can enable informed decision-making and provide an optimisation platform for the underlying system. By leveraging principles of Dynamic Data-Driven Applications Systems (DDDAS), DDTs can formulate computational modalities for feedback loops, model updates and decision-making, including autonomous ones. However, understanding autonomous decision-making often requires technical and domain-specific knowledge. This paper explores using large language models (LLMs) to provide an explainability platform for DDTs, generating natural language explanations of the system's decision-making by leveraging domain-specific knowledge bases. A case study from smart agriculture is presented.

Zongyou Yu, Qiang Qu, Qian Zhang et al.

Recent advancements in event-based recognition have demonstrated significant promise, yet most existing approaches rely on extensive training, limiting their adaptability for efficient processing of event-driven visual content. Meanwhile, large language models (LLMs) have exhibited remarkable zero-shot capabilities across diverse domains, but their application to event-based visual recognition remains largely unexplored. To bridge this gap, we propose \textbf{LLM-EvGen}, an event representation generator that produces LLM-compatible event representations \textbf{LLM-EvRep}, thereby enhancing the performance of LLMs on event recognition tasks. The generator is trained using a self-supervised framework, aligning the generated representations with semantic consistency and structural fidelity. Comprehensive experiments were conducted on three datasets: N-ImageNet, N-Caltech101, and N-MNIST. The results demonstrate that our method, \textbf{LLM-EvRep}, outperforms the event-to-video method, E2VID, by 15.93\%, 0.82\%, and 50.21\%, respectively, in recognition tasks when evaluated using GPT-4o.

Ming Li, Nan Zhang, Chenrui Fan et al.

While Large Reasoning Models (LRMs) generate extensive chain-of-thought reasoning, we lack a principled framework for understanding how these thoughts are structured. In this paper, we introduce a novel approach by applying Schoenfeld's Episode Theory, a classic cognitive framework for human mathematical problem-solving, to analyze the reasoning traces of LRMs. We annotated thousands of sentences and paragraphs from model-generated solutions to math problems using seven cognitive labels (e.g., Plan, Implement, Verify). The result is the first publicly available benchmark for the fine-grained analysis of machine reasoning, including a large annotated corpus and detailed annotation guidebooks. Our preliminary analysis reveals distinct patterns in LRM reasoning, such as the transition dynamics between cognitive states. This framework provides a theoretically grounded methodology for interpreting LRM cognition and enables future work on more controllable and transparent reasoning systems.

Priyanka Kargupta, Nan Zhang, Yunyi Zhang et al. · amazon-science

The rapid evolution of scientific fields introduces challenges in organizing and retrieving scientific literature. While expert-curated taxonomies have traditionally addressed this need, the process is time-consuming and expensive. Furthermore, recent automatic taxonomy construction methods either (1) over-rely on a specific corpus, sacrificing generalizability, or (2) depend heavily on the general knowledge of large language models (LLMs) contained within their pre-training datasets, often overlooking the dynamic nature of evolving scientific domains. Additionally, these approaches fail to account for the multi-faceted nature of scientific literature, where a single research paper may contribute to multiple dimensions (e.g., methodology, new tasks, evaluation metrics, benchmarks). To address these gaps, we propose TaxoAdapt, a framework that dynamically adapts an LLM-generated taxonomy to a given corpus across multiple dimensions. TaxoAdapt performs iterative hierarchical classification, expanding both the taxonomy width and depth based on corpus' topical distribution. We demonstrate its state-of-the-art performance across a diverse set of computer science conferences over the years to showcase its ability to structure and capture the evolution of scientific fields. As a multidimensional method, TaxoAdapt generates taxonomies that are 26.51% more granularity-preserving and 50.41% more coherent than the most competitive baselines judged by LLMs.

Yujia Chen, Changsong Li, Yiming Wang et al.

Mental health issues are worsening in today's competitive society, such as depression and anxiety. Traditional healings like counseling and chatbots fail to engage effectively, they often provide generic responses lacking emotional depth. Although large language models (LLMs) have the potential to create more human-like interactions, they still struggle to capture subtle emotions. This requires LLMs to be equipped with human-like adaptability and warmth. To fill this gap, we propose the MIND (Multi-agent INner Dialogue), a novel paradigm that provides more immersive psychological healing environments. Considering the strong generative and role-playing ability of LLM agents, we predefine an interactive healing framework and assign LLM agents different roles within the framework to engage in interactive inner dialogues with users, thereby providing an immersive healing experience. We conduct extensive human experiments in various real-world healing dimensions, and find that MIND provides a more user-friendly experience than traditional paradigms. This demonstrates that MIND effectively leverages the significant potential of LLMs in psychological healing.

Junjie Li, Nan Zhang, Xiaoyang Qu et al.

Object Navigation (ObjectNav) is a fundamental task in embodied artificial intelligence. Although significant progress has been made in semantic map construction and target direction prediction in current research, redundant exploration and exploration failures remain inevitable. A critical but underexplored direction is the timely termination of exploration to overcome these challenges. We observe a diminishing marginal effect between exploration steps and exploration rates and analyze the cost-benefit relationship of exploration. Inspired by this, we propose RATE-Nav, a Region-Aware Termination-Enhanced method. It includes a geometric predictive region segmentation algorithm and region-Based exploration estimation algorithm for exploration rate calculation. By leveraging the visual question answering capabilities of visual language models (VLMs) and exploration rates enables efficient termination.RATE-Nav achieves a success rate of 67.8% and an SPL of 31.3% on the HM3D dataset. And on the more challenging MP3D dataset, RATE-Nav shows approximately 10% improvement over previous zero-shot methods.

Nan Zhang, Eugene Kwek, Yusen Zhang et al.

Compression methods, including quantization, distillation, and pruning, improve the computational efficiency of large reasoning models (LRMs). However, existing studies either fail to sufficiently compare all three compression methods on LRMs or lack in-depth interpretation analysis. In this paper, we investigate how the reasoning capabilities of LRMs are compromised during compression, through performance benchmarking and mechanistic interpretation. To uncover the effects of compression on reasoning performance, we benchmark quantized, distilled, and pruned DeepSeek-R1 models on four reasoning datasets (AIME 2024, FOLIO, Temporal Sequences, and MuSiQue). To precisely locate compression effects on model weights, we adapt difference of means and attribution patching techniques, focusing on the activation of every linear component in compressed LRMs, to interpret fine-grained causal relationships between weights and various reasoning capabilities. This fine-grained interpretation addresses a fundamental question of compression: which weights are the most important for reasoning? Overall, we find dynamically quantized 2.51-bit R1 reaches close-to-R1 performance. With empirical verification, we present three main findings that generalize across both Llama and Qwen: (1) Weight count has a greater impact on LRMs' knowledge memorization than reasoning, highlighting the risks of pruning and distillation; (2) The MLP up projection in the final layer of distilled LRMs is one of the most important components, offering a new perspective on locating critical weights - a fundamental problem in model compression; and (3) Current quantization methods overly compress the final-layer modules and MLP gate projections, so protecting just 2% of all weights that are excessively compressed can raise average accuracy by 6.57%, greatly surpassing the state-of-the-art.

Yizheng Wang, Tao Jiang, Jinyan Bai et al.

Software Requirement Document (RD) typically contain tens of thousands of individual requirements, and ensuring consistency among these requirements is critical for the success of software engineering projects. Automated detection methods can significantly enhance efficiency and reduce costs; however, existing approaches still face several challenges, including low detection accuracy on imbalanced data, limited semantic extraction due to the use of a single encoder, and suboptimal performance in cross-domain transfer learning. To address these issues, this paper proposes a Transferable Software Requirement Conflict Detection Framework based on SBERT and SimCSE, termed TSRCDF-SS. First, the framework employs two independent encoders, Sentence-BERT (SBERT) and Simple Contrastive Sentence Embedding (SimCSE), to generate sentence embeddings for requirement pairs, followed by a six-element concatenation strategy. Furthermore, the classifier is enhanced by a two-layer fully connected feedforward neural network (FFNN) with a hybrid loss optimization strategy that integrates a variant of Focal Loss, domain-specific constraints, and a confidence-based penalty term. Finally, the framework synergistically integrates sequential and cross-domain transfer learning. Experimental results demonstrate that the proposed framework achieves a 10.4% improvement in both macro-F1 and weighted-F1 scores in in-domain settings, and an 11.4% increase in macro-F1 in cross-domain scenarios.

Qingshu Xu, Hong Jiao, Tianyi Zhou et al.

Accurate alignment of items to content standards is critical for valid score interpretation in large-scale assessments. This study evaluates three automated paradigms for aligning items with four domain and nineteen skill labels. First, we extracted embeddings and trained multiple classical supervised machine learning models, and further investigated the impact of dimensionality reduction on model performance. Second, we fine-tuned eight BERT model and its variants for both domain and skill alignment. Third, we explored ensemble learning with majority voting and stacking with multiple meta-models. The DeBERTa-v3-base achieved the highest weighted-average F1 score of 0.950 for domain alignment while the RoBERTa-large yielded the highest F1 score of 0.869 for skill alignment. Ensemble models did not surpass the best-performing language models. Dimension reduction enhanced linear classifiers based on embeddings but did not perform better than language models. This study demonstrated different methods in automated item alignment to content standards.}

Yanbin Fu, Hong Jiao, Tianyi Zhou et al.

Aligning test items to content standards is a critical step in test development to collect validity evidence based on content. Item alignment has typically been conducted by human experts. This judgmental process can be subjective and time-consuming. This study investigated the performance of fine-tuned small language models (SLMs) for automated item alignment using data from a large-scale standardized reading and writing test for college admissions. Different SLMs were trained for alignment at both domain and skill levels respectively with 10 skills mapped to 4 content domains. The model performance was evaluated in multiple criteria on two testing datasets. The impact of types and sizes of the input data for training was investigated. Results showed that including more item text data led to substantially better model performance, surpassing the improvements induced by sample size increase alone. For comparison, supervised machine learning models were trained using the embeddings from the multilingual-E5-large-instruct model. The study results showed that fine-tuned SLMs consistently outperformed the embedding-based supervised machine learning models, particularly for the more fine-grained skill alignment. To better understand model misclassifications, multiple semantic similarity analysis including pairwise cosine similarity, Kullback-Leibler divergence of embedding distributions, and two-dimension projections of item embeddings were conducted. These analyses consistently showed that certain skills in SAT and PSAT were semantically too close, providing evidence for the observed misclassification.

Sydney Peters, Nan Zhang, Hong Jiao et al.

Item difficulty plays a crucial role in test performance, interpretability of scores, and equity for all test-takers, especially in large-scale assessments. Traditional approaches to item difficulty modeling rely on field testing and classical test theory (CTT)-based item analysis or item response theory (IRT) calibration, which can be time-consuming and costly. To overcome these challenges, text-based approaches leveraging machine learning and language models, have emerged as promising alternatives. This paper reviews and synthesizes 37 articles on automated item difficulty prediction in large-scale assessment settings published through May 2025. For each study, we delineate the dataset, difficulty parameter, subject domain, item type, number of items, training and test data split, input, features, model, evaluation criteria, and model performance outcomes. Results showed that although classic machine learning models remain relevant due to their interpretability, state-of-the-art language models, using both small and large transformer-based architectures, can capture syntactic and semantic patterns without the need for manual feature engineering. Uniquely, model performance outcomes were summarized to serve as a benchmark for future research and overall, text-based methods have the potential to predict item difficulty with root mean square error (RMSE) as low as 0.165, Pearson correlation as high as 0.87, and accuracy as high as 0.806. The review concludes by discussing implications for practice and outlining future research directions for automated item difficulty modeling.

Haojun Yu, Youcheng Li, Zihan Niu et al.

Breast ultrasound (BUS) is an essential tool for diagnosing breast lesions, with millions of examinations per year. However, publicly available high-quality BUS benchmarks for AI development are limited in data scale and annotation richness. In this work, we present BUS-CoT, a BUS dataset for chain-of-thought (CoT) reasoning analysis, which contains 11,439 images of 10,019 lesions from 4,838 patients and covers all 99 histopathology types. To facilitate research on incentivizing CoT reasoning, we construct the reasoning processes based on observation, feature, diagnosis and pathology labels, annotated and verified by experienced experts. Moreover, by covering lesions of all histopathology types, we aim to facilitate robust AI systems in rare cases, which can be error-prone in clinical practice.

Yunqing Liu, Nan Zhang, Zhiming Tan

Effective specification-aware part retrieval within complex CAD assemblies is essential for automated design verification and downstream engineering tasks. However, directly using LLMs/VLMs to this task presents some challenges: the input sequences may exceed model token limits, and even after processing, performance remains unsatisfactory. Moreover, fine-tuning LLMs/VLMs requires significant computational resources, and for many high-performing general-use proprietary models (e.g., GPT or Gemini), fine-tuning access is not available. In this paper, we propose a novel part retrieval framework that requires no extra training, but using Error Notebooks + RAG for refined prompt engineering to help improve the existing general model's retrieval performance. The construction of Error Notebooks consists of two steps: (1) collecting historical erroneous CoTs and their incorrect answers, and (2) connecting these CoTs through reflective corrections until the correct solutions are obtained. As a result, the Error Notebooks serve as a repository of tasks along with their corrected CoTs and final answers. RAG is then employed to retrieve specification-relevant records from the Error Notebooks and incorporate them into the inference process. Another major contribution of our work is a human-in-the-loop CAD dataset, which is used to evaluate our method. In addition, the engineering value of our novel framework lies in its ability to effectively handle 3D models with lengthy, non-natural language metadata. Experiments with proprietary models, including GPT-4o and the Gemini series, show substantial gains, with GPT-4o (Omni) achieving up to a 23.4% absolute accuracy improvement on the human preference dataset. Moreover, ablation studies confirm that CoT reasoning provides benefits especially in challenging cases with higher part counts (>10).

Chenyu Lu, Shiliang Sun, Jing Zhao et al.

Multimodal machine translation (MMT) seeks to address the challenges posed by linguistic polysemy and ambiguity in translation tasks by incorporating visual information. A key bottleneck in current MMT research is the effective utilization of visual data. Previous approaches have focused on extracting global or region-level image features and using attention or gating mechanisms for multimodal information fusion. However, these methods have not adequately tackled the issue of visual information redundancy in MMT, nor have they proposed effective solutions. In this paper, we introduce a novel approach--multimodal machine translation with visual Scene Graph Pruning (PSG), which leverages language scene graph information to guide the pruning of redundant nodes in visual scene graphs, thereby reducing noise in downstream translation tasks. Through extensive comparative experiments with state-of-the-art methods and ablation studies, we demonstrate the effectiveness of the PSG model. Our results also highlight the promising potential of visual information pruning in advancing the field of MMT.

Yusen Zhang, Wenliang Zheng, Aashrith Madasu et al.

High-resolution image (HRI) understanding aims to process images with a large number of pixels, such as pathological images and agricultural aerial images, both of which can exceed 1 million pixels. Vision Large Language Models (VLMs) can allegedly handle HRIs, however, there is a lack of a comprehensive benchmark for VLMs to evaluate HRI understanding. To address this gap, we introduce HRScene, a novel unified benchmark for HRI understanding with rich scenes. HRScene incorporates 25 real-world datasets and 2 synthetic diagnostic datasets with resolutions ranging from 1,024 $\times$ 1,024 to 35,503 $\times$ 26,627. HRScene is collected and re-annotated by 10 graduate-level annotators, covering 25 scenarios, ranging from microscopic to radiology images, street views, long-range pictures, and telescope images. It includes HRIs of real-world objects, scanned documents, and composite multi-image. The two diagnostic evaluation datasets are synthesized by combining the target image with the gold answer and distracting images in different orders, assessing how well models utilize regions in HRI. We conduct extensive experiments involving 28 VLMs, including Gemini 2.0 Flash and GPT-4o. Experiments on HRScene show that current VLMs achieve an average accuracy of around 50% on real-world tasks, revealing significant gaps in HRI understanding. Results on synthetic datasets reveal that VLMs struggle to effectively utilize HRI regions, showing significant Regional Divergence and lost-in-middle, shedding light on future research.