Xinyu Zhao

Zayne Sprague, Fangcong Yin, Juan Diego Rodriguez et al.

Chain-of-thought (CoT) via prompting is the de facto method for eliciting reasoning capabilities from large language models (LLMs). But for what kinds of tasks is this extra ``thinking'' really helpful? To analyze this, we conducted a quantitative meta-analysis covering over 100 papers using CoT and ran our own evaluations of 20 datasets across 14 models. Our results show that CoT gives strong performance benefits primarily on tasks involving math or logic, with much smaller gains on other types of tasks. On MMLU, directly generating the answer without CoT leads to almost identical accuracy as CoT unless the question or model's response contains an equals sign, indicating symbolic operations and reasoning. Following this finding, we analyze the behavior of CoT on these problems by separating planning and execution and comparing against tool-augmented LLMs. Much of CoT's gain comes from improving symbolic execution, but it underperforms relative to using a symbolic solver. Our results indicate that CoT can be applied selectively, maintaining performance while saving inference costs. Furthermore, they suggest a need to move beyond prompt-based CoT to new paradigms that better leverage intermediate computation across the whole range of LLM applications.

Manya Wadhwa, Xinyu Zhao, Junyi Jessy Li et al.

Recent work has explored the capability of large language models (LLMs) to identify and correct errors in LLM-generated responses. These refinement approaches frequently evaluate what sizes of models are able to do refinement for what problems, but less attention is paid to what effective feedback for refinement looks like. In this work, we propose looking at refinement with feedback as a composition of three distinct LLM competencies: (1) detection of bad generations; (2) fine-grained natural language critique generation; (3) refining with fine-grained feedback. The first step can be implemented with a high-performing discriminative model and steps 2 and 3 can be implemented either via prompted or fine-tuned LLMs. A key property of the proposed Detect, Critique, Refine ("DCR") method is that the step 2 critique model can give fine-grained feedback about errors, made possible by offloading the discrimination to a separate model in step 1. We show that models of different capabilities benefit from refining with DCR on the task of improving factual consistency of document grounded summaries. Overall, DCR consistently outperforms existing end-to-end refinement approaches and current trained models not fine-tuned for factuality critiquing.

Xinyu Zhao, Qingyun Sun, Jiayi Luo et al.

Zero-Shot Anomaly Detection (ZSAD) aims to detect anomalies in unseen domains without target-domain adaptation. Recent CLIP-based methods have shown promising performance by leveraging prompt learning and visual-text alignment. However, most existing approaches rely on a single adaptation pathway, which may be insufficient for heterogeneous anomaly patterns across domains. In practice, anomalies exhibit vastly different characteristics, ranging from salient, localized structural disruptions to subtle, diffuse, and irregular variations. To address this challenge, we propose EntroAD, a structural entropy-guided zero-shot anomaly detection framework. Unlike previous methods, EntroAD introduces a dynamic routing mechanism to process different types of anomalies with specialized adaptation strategies. Specifically, we estimate patch-level structural entropy from self-attention-induced patch relations and use it as a proxy for relational uncertainty to guide anomaly-aware token routing. Based on this routing signal, we construct anomaly-aware routed tokens to better capture anomaly cues with different structural characteristics. We further introduce a confidence-aware dual-branch prompt adaptation module to stabilize visual-text alignment while preserving CLIP's transferable prior. Extensive experiments on 10 industrial and medical benchmarks show that EntroAD achieves state-of-the-art performance in challenging cross-dataset ZSAD settings.

Zhen Tan, Daize Dong, Xinyu Zhao et al.

In this paper, we introduce Dynamic Layer Operations (DLO), a novel approach for vertically scaling transformer-based Large Language Models (LLMs) by dynamically expanding, activating, or skipping layers using a sophisticated routing policy based on layerwise feature similarity. Unlike traditional Mixture-of-Experts (MoE) methods that focus on extending the model width, our approach targets model depth, addressing the redundancy observed across layer representations for various input samples. Our framework is integrated with the Supervised Fine-Tuning (SFT) stage, eliminating the need for resource-intensive Continual Pre-Training (CPT). Experimental results demonstrate that DLO not only outperforms the original unscaled models but also achieves comparable results to densely expanded models with significantly improved efficiency. Our work offers a promising direction for building efficient yet powerful LLMs. We will release our implementation and model weights upon acceptance.

Yuhan Wang, Haopeng Zhang, Yibo Ding et al.

Pre-training on text-attributed graphs (TAGs) is central to building transferable graph foundation models, where LLM-as-Aligner methods align graph and text representations through the semantic knowledge of large language models. However, these methods usually assume that node texts provide sufficient and reliable supervision, an assumption often violated in real-world sparse TAGs. When textual anchors are missing, noisy, or uneven across domains, graph structures must be aligned with weak semantic evidence, leading to unreliable structure-semantics correspondence and sparsity-induced transfer bias. This paper presents S2Aligner, a sparsity-aware and structure-enhanced LLM-as-Aligner framework for graph-text pre-training on sparse TAGs. The key idea is to decouple semantic alignment from structural modeling, allowing topology-aware signals to enhance alignment without contaminating the shared semantic space. Specifically, S2Aligner decomposes graph-text representations into semantic and structural components, uses structure-oriented reconstruction with consistency control to inject reliable topology cues into text representations, and suppresses inconsistent structural signals under textual sparsity. Moreover, S2Aligner introduces sparsity-aware cross-domain risk balancing, which calibrates domain risks through a global-domain density ratio and downweights unreliable sparse samples via graph reliability estimation. Theoretical analysis shows that this objective reduces cross-domain generalization gaps by controlling domain risk discrepancy. Extensive experiments across diverse graph domains, sparsity levels, and downstream tasks demonstrate that S2Aligner consistently outperforms existing baselines.

Bernardo Consoli, Xizhi Wu, Song Wang et al.

Extracting social determinants of health (SDoH) from unstructured medical notes depends heavily on labor-intensive annotations, which are typically task-specific, hampering reusability and limiting sharing. In this study we introduced SDoH-GPT, a simple and effective few-shot Large Language Model (LLM) method leveraging contrastive examples and concise instructions to extract SDoH without relying on extensive medical annotations or costly human intervention. It achieved tenfold and twentyfold reductions in time and cost respectively, and superior consistency with human annotators measured by Cohen's kappa of up to 0.92. The innovative combination of SDoH-GPT and XGBoost leverages the strengths of both, ensuring high accuracy and computational efficiency while consistently maintaining 0.90+ AUROC scores. Testing across three distinct datasets has confirmed its robustness and accuracy. This study highlights the potential of leveraging LLMs to revolutionize medical note classification, demonstrating their capability to achieve highly accurate classifications with significantly reduced time and cost.

Xinyu Zhao, Qingyun Sun, Jiayi Luo et al.

Graph Anomaly Detection (GAD) aims to identify irregular patterns in graph data, and recent works have explored zero-shot generalist GAD to enable generalization to unseen graph datasets. However, existing zero-shot GAD methods largely ignore intrinsic geometric differences across diverse anomaly patterns, substantially limiting their cross-domain generalization. In this work, we reveal that anomaly detectability is highly dependent on the underlying geometric properties and that embedding graphs from different domains into a single static curvature space can distort the structural signatures of anomalies. To address the challenge that a single curvature space cannot capture geometry-dependent graph anomaly patterns, we propose GAD-MoRE, a novel framework for zero-shot Generalizable Graph Anomaly Detection with a Mixture of Riemannian Experts architecture. Specifically, to ensure that each anomaly pattern is modeled in the Riemannian space where it is most detectable, GAD-MoRE employs a set of specialized Riemannian expert networks, each operating in a distinct curvature space. To align raw node features with curvature-specific anomaly characteristics, we introduce an anomaly-aware multi-curvature feature alignment module that projects inputs into parallel Riemannian spaces, enabling the capture of diverse geometric characteristics. Finally, to facilitate better generalization beyond seen patterns, we design a memory-based dynamic router that adaptively assigns each input to the most compatible expert based on historical reconstruction performance on similar anomalies. Extensive experiments in the zero-shot setting demonstrate that GAD-MoRE significantly outperforms state-of-the-art generalist GAD baselines, and even surpasses strong competitors that are few-shot fine-tuned with labeled data from the target domain.

Liyan Tang, Grace Kim, Xinyu Zhao et al.

Chart understanding presents a unique challenge for large vision-language models (LVLMs), as it requires the integration of sophisticated textual and visual reasoning capabilities. However, current LVLMs exhibit a notable imbalance between these skills, falling short on visual reasoning that is difficult to perform in text. We conduct a case study using a synthetic dataset solvable only through visual reasoning and show that model performance degrades significantly with increasing visual complexity, while human performance remains robust. We then introduce ChartMuseum, a new Chart Question Answering (QA) benchmark containing 1,162 expert-annotated questions spanning multiple reasoning types, curated from real-world charts across 184 sources, specifically built to evaluate complex visual and textual reasoning. Unlike prior chart understanding benchmarks -- where frontier models perform similarly and near saturation -- our benchmark exposes a substantial gap between model and human performance, while effectively differentiating model capabilities: although humans achieve 93% accuracy, the best-performing model Gemini-2.5-Pro attains only 63.0%, and the leading open-source LVLM Qwen2.5-VL-72B-Instruct achieves only 38.5%. Moreover, on questions requiring primarily visual reasoning, all models experience a 35%-55% performance drop from text-reasoning-heavy question performance. Lastly, our qualitative error analysis reveals specific categories of visual reasoning that are challenging for current LVLMs.

Guanjie Chen, Xinyu Zhao, Yucheng Zhou et al.

Diffusion Transformers (DiT) have emerged as a powerful architecture for image and video generation, offering superior quality and scalability. However, their practical application suffers from inherent dynamic feature instability, leading to error amplification during cached inference. Through systematic analysis, we identify the absence of long-range feature preservation mechanisms as the root cause of unstable feature propagation and perturbation sensitivity. To this end, we propose Skip-DiT, an image and video generative DiT variant enhanced with Long-Skip-Connections (LSCs) - the key efficiency component in U-Nets. Theoretical spectral norm and visualization analysis demonstrate how LSCs stabilize feature dynamics. Skip-DiT architecture and its stabilized dynamic feature enable an efficient statical caching mechanism that reuses deep features across timesteps while updating shallow components. Extensive experiments across the image and video generation tasks demonstrate that Skip-DiT achieves: (1) 4.4 times training acceleration and faster convergence, (2) 1.5-2 times inference acceleration with negligible quality loss and high fidelity to the original output, outperforming existing DiT caching methods across various quantitative metrics. Our findings establish Long-Skip-Connections as critical architectural components for stable and efficient diffusion transformers. Codes are provided in the https://github.com/OpenSparseLLMs/Skip-DiT.

Kaixin zhang, Xiaohe Li, Jiahao Li et al.

Multi-modal Large Language Models (MLLMs) have significantly advanced video reasoning, yet Video Question Answering (VideoQA) remains challenging due to its demand for temporal causal reasoning and evidence-grounded answer generation. Prevailing end-to-end MLLM frameworks lack explicit structured reasoning between visual perception and answer derivation, causing severe hallucinations and poor interpretability. Existing methods also fail to address three core gaps: faithful visual clue extraction, utility-aware clue filtering, and end-to-end clue-answer alignment. Inspired by hierarchical human visual cognition, we propose ClueNet, a clue-aware video reasoning framework with a two-stage supervised fine-tuning paradigm without extensive base model modifications. Decoupled supervision aligns clue extraction and chain-based reasoning, while inference supervision with an adaptive clue filter refines high-order reasoning, alongside lightweight modules for efficient inference. Experiments on NExT-QA, STAR, and MVBench show that ClueNet outperforms state-of-the-art methods by $\ge$ 1.1%, with superior generalization, hallucination mitigation, inference efficiency, and cross-backbone compatibility. This work bridges the perception-to-generation gap in MLLM video understanding, providing an interpretable, faithful reasoning paradigm for high-stakes VideoQA applications.

ChengYou Li, XiaoDong Liu, XiangBao Meng et al.

The paradigm of Large Language Models is undergoing a fundamental transition from static inference engines to dynamic autonomous cognitive systems.While current research primarily focuses on scaling context windows or optimizing prompt engineering the theoretical bridge between micro scale token processing and macro scale systemic intelligence remains fragmented.This paper proposes AgentOS,a holistic conceptual framework that redefines the LLM as a "Reasoning Kernel" governed by structured operating system logic.Central to this architecture is Deep Context Management which conceptualizes the context window as an Addressable Semantic Space rather than a passive buffer.We systematically deconstruct the transition from discrete sequences to coherent cognitive states introducing mechanisms for Semantic Slicing and Temporal Alignment to mitigate cognitive drift in multi-agent orchestration.By mapping classical OS abstractions such as memory paging interrupt handling and process scheduling onto LLM native constructs, this review provides a rigorous roadmap for architecting resilient scalable and self-evolving cognitive environments.Our analysis asserts that the next frontier of AGI development lies in the architectural efficiency of system-level coordination.

Guanjie Chen, Xinyu Zhao, Tianlong Chen et al.

Mixture-of-Experts (MoE) has gained increasing popularity as a promising framework for scaling up large language models (LLMs). However, the reliability assessment of MoE lags behind its surging applications. Moreover, when transferred to new domains such as in fine-tuning MoE models sometimes underperform their dense counterparts. Motivated by the research gap and counter-intuitive phenomenon, we propose $\texttt{MoE-RBench}$, the first comprehensive assessment of SMoE reliability from three aspects: $\textit{(i)}$ safety and hallucination, $\textit{(ii)}$ resilience to adversarial attacks, and $\textit{(iii)}$ out-of-distribution robustness. Extensive models and datasets are tested to compare the MoE to dense networks from these reliability dimensions. Our empirical observations suggest that with appropriate hyperparameters, training recipes, and inference techniques, we can build the MoE model more reliably than the dense LLM. In particular, we find that the robustness of SMoE is sensitive to the basic training settings. We hope that this study can provide deeper insights into how to adapt the pre-trained MoE model to other tasks with higher-generation security, quality, and stability. Codes are available at https://github.com/UNITES-Lab/MoE-RBench

Song Wang, Zhenyu Lei, Zhen Tan et al.

Functional Magnetic Resonance Image (fMRI) is commonly employed to study human brain activity, since it offers insight into the relationship between functional fluctuations and human behavior. To enhance analysis and comprehension of brain activity, Graph Neural Networks (GNNs) have been widely applied to the analysis of functional connectivities (FC) derived from fMRI data, due to their ability to capture the synergistic interactions among brain regions. However, in the human brain, performing complex tasks typically involves the activation of certain pathways, which could be represented as paths across graphs. As such, conventional GNNs struggle to learn from these pathways due to the long-range dependencies of multiple pathways. To address these challenges, we introduce a novel framework BrainMAP to learn Multiple Activation Pathways in Brain networks. BrainMAP leverages sequential models to identify long-range correlations among sequentialized brain regions and incorporates an aggregation module based on Mixture of Experts (MoE) to learn from multiple pathways. Our comprehensive experiments highlight BrainMAP's superior performance. Furthermore, our framework enables explanatory analyses of crucial brain regions involved in tasks. Our code is provided at https://github.com/LzyFischer/Graph-Mamba.

Xinyu Zhao, Nikita Karagodin, Hamed Hassani et al.

While many approaches to improve VQ-VAE performance focus on codebook size and utilization, the effect of dimensional collapse, where trained VQ-VAE representations live in an extremely low-dimensional subspace (1-2% of full rank), remains unaddressed. We show theoretically and empirically that dimension collapse causes a hard loss lower bound that various codebook improvement techniques fail to surpass. Our analytic framework extends the sequential learning effect of Saxe et al. [2014] by introducing ideas from rate-distortion theory and explains how the latent collapse is caused by the VQ suppressing lower-variance directions. Our theory justifies a simple solution: a "warm-up phase" that trains the model as an (unquantized) autoencoder before introducing VQ. On both synthetic experiments and large-scale image (VQGAN) and audio (WavTokenizer) VQ-VAEs, we show that AE Warm-Up successfully restores representation dimension, leading to lower reconstruction and perceptual loss at the same training budget. Across codebook sizes $K \in$ {$2^{10}, 2^{14}, 2^{16}$}, AE warm-up raises VQGAN codebook effective dimension from 3-5 to 17-19 and reduces rFID by 17-35%; on WavTokenizer at $K \in$ {$2^{13}, 2^{14}$}, it raises codebook dimension from 4 to 17-19 and improves PESQ by 11-14%. We empirically characterize how warm-up duration governs the achievable final loss. In agreement with experiment, our theoretical analysis predicts downstream performance as a function of warm-up length, enabling an adaptive criterion for switching from AE Warm-up to VQ-VAE training.

Bo Liu, Liqiang Yu, Chang Che et al.

This paper focuses on the analysis of the application effectiveness of the integration of deep learning and computer vision technologies. Deep learning achieves a historic breakthrough by constructing hierarchical neural networks, enabling end-to-end feature learning and semantic understanding of images. The successful experiences in the field of computer vision provide strong support for training deep learning algorithms. The tight integration of these two fields has given rise to a new generation of advanced computer vision systems, significantly surpassing traditional methods in tasks such as machine vision image classification and object detection. In this paper, typical image classification cases are combined to analyze the superior performance of deep neural network models while also pointing out their limitations in generalization and interpretability, proposing directions for future improvements. Overall, the efficient integration and development trend of deep learning with massive visual data will continue to drive technological breakthroughs and application expansion in the field of computer vision, making it possible to build truly intelligent machine vision systems. This deepening fusion paradigm will powerfully promote unprecedented tasks and functions in computer vision, providing stronger development momentum for related disciplines and industries.

Chang Che, Qunwei Lin, Xinyu Zhao et al.

The process of transforming input images into corresponding textual explanations stands as a crucial and complex endeavor within the domains of computer vision and natural language processing. In this paper, we propose an innovative ensemble approach that harnesses the capabilities of Contrastive Language-Image Pretraining models.

Liqiang Yu, Bo Liu, Qunwei Lin et al.

In the realm of patent document analysis, assessing semantic similarity between phrases presents a significant challenge, notably amplifying the inherent complexities of Cooperative Patent Classification (CPC) research. Firstly, this study addresses these challenges, recognizing early CPC work while acknowledging past struggles with language barriers and document intricacy. Secondly, it underscores the persisting difficulties of CPC research. To overcome these challenges and bolster the CPC system, This paper presents two key innovations. Firstly, it introduces an ensemble approach that incorporates four BERT-related models, enhancing semantic similarity accuracy through weighted averaging. Secondly, a novel text preprocessing method tailored for patent documents is introduced, featuring a distinctive input structure with token scoring that aids in capturing semantic relationships during CPC context training, utilizing BCELoss. Our experimental findings conclusively establish the effectiveness of both our Ensemble Model and novel text processing strategies when deployed on the U.S. Patent Phrase to Phrase Matching dataset.

Tianhao Fu, Xinxin Xu, Weichen Xu et al.

Market making (MM) through Reinforcement Learning (RL) has attracted significant attention in financial trading. With the development of Large Language Models (LLMs), more and more attempts are being made to apply LLMs to financial areas. A simple, direct application of LLM as an agent shows significant performance. Such methods are hindered by their slow inference speed, while most of the current research has not studied LLM distillation for this specific task. To address this, we first propose the normalized fluorescent probe to study the mechanism of the LLM's feature. Based on the observation found by our investigation, we propose Cooperative Market Making (CMM), a novel framework that decouples LLM features across three orthogonal dimensions: layer, task, and data. Various student models collaboratively learn simple LLM features along with different dimensions, with each model responsible for a distinct feature to achieve knowledge distillation. Furthermore, CMM introduces an Hájek-MoE to integrate the output of the student models by investigating the contribution of different models in a kernel function-generated common feature space. Extensive experimental results on four real-world market datasets demonstrate the superiority of CMM over the current distillation method and RL-based market-making strategies.

Xiang Ao, Jingxuan Zhang, Xinyu Zhao

Accurately predicting stock repurchases is crucial for quantitative investment and risk management, yet traditional static models fail to capture the complex temporal dependencies of corporate financial conditions. This paper proposes a dynamic early warning system integrating economic theory with deep temporal networks. Using Chinese A-share panel data (2014-2024), we employ a hybrid Temporal Convolutional Network (TCN) and Attention-based LSTM to capture long- and short-term financial evolutionary patterns. Rolling-window cross-validation demonstrates our model significantly outperforms static baselines like Logistic Regression and XGBoost. Furthermore, utilizing Explainable AI (XAI), we reveal the temporal dynamics of repurchase decisions: prolonged "undervaluation" serves as the long-term underlying motive, while a sharp increase in "cash flow" acts as the decisive short-term trigger. This study provides a robust deep learning paradigm for financial forecasting and offers dynamic empirical support for classic corporate finance hypotheses.

Jiaxin Huang, Xinyu Zhao, Chang Che et al.

The objective of this study is to improve automated feedback tools designed for English Language Learners (ELLs) through the utilization of data science techniques encompassing machine learning, natural language processing, and educational data analytics. Automated essay scoring (AES) research has made strides in evaluating written essays, but it often overlooks the specific needs of English Language Learners (ELLs) in language development. This study explores the application of BERT-related techniques to enhance the assessment of ELLs' writing proficiency within AES. To address the specific needs of ELLs, we propose the use of DeBERTa, a state-of-the-art neural language model, for improving automated feedback tools. DeBERTa, pretrained on large text corpora using self-supervised learning, learns universal language representations adaptable to various natural language understanding tasks. The model incorporates several innovative techniques, including adversarial training through Adversarial Weights Perturbation (AWP) and Metric-specific AttentionPooling (6 kinds of AP) for each label in the competition. The primary focus of this research is to investigate the impact of hyperparameters, particularly the adversarial learning rate, on the performance of the model. By fine-tuning the hyperparameter tuning process, including the influence of 6AP and AWP, the resulting models can provide more accurate evaluations of language proficiency and support tailored learning tasks for ELLs. This work has the potential to significantly benefit ELLs by improving their English language proficiency and facilitating their educational journey.

Xinyu Zhao, Fangcong Yin, Greg Durrett

Long-context LLMs are increasingly in demand for applications such as retrieval-augmented generation. To defray the cost of pretraining LLMs over long contexts, recent work takes an approach of synthetic context extension: fine-tuning LLMs with synthetically generated long-context data in a post-training stage. However, it remains unclear how and why this synthetic context extension imparts abilities for downstream long-context tasks. In this paper, we investigate fine-tuning on synthetic data for three long-context tasks that require retrieval and reasoning. We vary the realism of "needle" concepts to be retrieved and diversity of the surrounding "haystack" context, from using LLMs to construct synthetic documents to using templated relations and creating symbolic datasets. We find that models trained on synthetic data fall short of the real data, but surprisingly, the mismatch can be interpreted and even predicted in terms of a special set of attention heads that are responsible for retrieval over long context, retrieval heads (Wu et al., 2024). The retrieval heads learned on synthetic data have high overlap with retrieval heads learned on real data, and there is a strong correlation between the recall of heads learned and the downstream performance of a model. Furthermore, with attention knockout and activation patching, we mechanistically show that retrieval heads are necessary and explain model performance, although they are not totally sufficient. Our results shed light on how to interpret synthetic data fine-tuning performance and how to approach creating better data for learning real-world capabilities over long contexts.

Xinyu Zhao, Hao Yan, Yongming Liu

A large volume of accident reports is recorded in the aviation domain, which greatly values improving aviation safety. To better use those reports, we need to understand the most important events or impact factors according to the accident reports. However, the increasing number of accident reports requires large efforts from domain experts to label those reports. In order to make the labeling process more efficient, many researchers have started developing algorithms to identify the underlying events from accident reports automatically. This article argues that we can identify the events more accurately by leveraging the event taxonomy. More specifically, we consider the problem a hierarchical classification task where we first identify the coarse-level information and then predict the fine-level information. We achieve this hierarchical classification process by incorporating a novel hierarchical attention module into BERT. To further utilize the information from event taxonomy, we regularize the proposed model according to the relationship and distribution among labels. The effectiveness of our framework is evaluated with the data collected by National Transportation Safety Board (NTSB). It has been shown that fine-level prediction accuracy is highly improved, and the regularization term can be beneficial to the rare event identification problem.

Xiaohe Li, Haohua Wu, Jiahao Li et al.

The rapid increase in remote sensing satellites has led to the emergence of distributed space-based observation systems. However, existing distributed remote sensing models often rely on centralized training, resulting in data leakage, communication overhead, and reduced accuracy due to data distribution discrepancies across platforms. To address these challenges, we propose the \textit{Self-Adjustment FEderated Learning} (SAFE) framework, which innovatively leverages federated learning to enhance collaborative sensing in remote sensing scenarios. SAFE introduces four key strategies: (1) \textit{Class Rectification Optimization}, which autonomously addresses class imbalance under unknown local and global distributions. (2) \textit{Feature Alignment Update}, which mitigates Non-IID data issues via locally controlled EMA updates. (3) \textit{Dual-Factor Modulation Rheostat}, which dynamically balances optimization effects during training. (4) \textit{Adaptive Context Enhancement}, which is designed to improve model performance by dynamically refining foreground regions, ensuring computational efficiency with accuracy improvement across distributed satellites. Experiments on real-world image classification and object segmentation datasets validate the effectiveness and reliability of the SAFE framework in complex remote sensing scenarios.

XinYu Zhao, ChengYou Li, XiangBao Meng et al.

Generative Engine Optimization (GEO) is rapidly reshaping digital marketing paradigms in the era of Large Language Models (LLMs). However, current GEO strategies predominantly rely on Retrieval-Augmented Generation (RAG), which inherently suffers from probabilistic hallucinations and the "zero-click" paradox, failing to establish sustainable commercial trust. In this paper, we systematically deconstruct the probabilistic flaws of existing RAG-based GEO and propose a paradigm shift towards deterministic multi-agent intent routing. First, we mathematically formulate Semantic Entropy Drift (SED) to model the dynamic decay of confidence curves in LLMs over continuous temporal and contextual perturbations. To rigorously quantify optimization value in black-box commercial engines, we introduce the Isomorphic Attribution Regression (IAR) model, leveraging a Multi-Agent System (MAS) probe with strict human-in-the-loop physical isolation to enforce hallucination penalties. Furthermore, we architect the Deterministic Agent Handoff (DAH) protocol, conceptualizing an Agentic Trust Brokerage (ATB) ecosystem where LLMs function solely as intent routers rather than final answer generators. We empirically validate this architecture using EasyNote, an industrial AI meeting minutes product by Yishu Technology. By routing the intent of "knowledge graph mapping on an infinite canvas" directly to its specialized proprietary agent via DAH, we demonstrate the reduction of vertical task hallucination rates to near zero. This work establishes a foundational theoretical framework for next-generation GEO and paves the way for a well-ordered, deterministic human-AI collaboration ecosystem.

Tianxiang Xu, Zhichao Wen, Xinyu Zhao et al.

The increasing integration of Industrial IoT (IIoT) exposes critical cyber-physical systems to sophisticated, multi-stage attacks that elude traditional defenses lacking contextual awareness. This paper introduces L2M-AID, a novel framework for Autonomous Industrial Defense using LLM-empowered, Multi-agent reinforcement learning. L2M-AID orchestrates a team of collaborative agents, each driven by a Large Language Model (LLM), to achieve adaptive and resilient security. The core innovation lies in the deep fusion of two AI paradigms: we leverage an LLM as a semantic bridge to translate vast, unstructured telemetry into a rich, contextual state representation, enabling agents to reason about adversary intent rather than merely matching patterns. This semantically-aware state empowers a Multi-Agent Reinforcement Learning (MARL) algorithm, MAPPO, to learn complex cooperative strategies. The MARL reward function is uniquely engineered to balance security objectives (threat neutralization) with operational imperatives, explicitly penalizing actions that disrupt physical process stability. To validate our approach, we conduct extensive experiments on the benchmark SWaT dataset and a novel synthetic dataset generated based on the MITRE ATT&CK for ICS framework. Results demonstrate that L2M-AID significantly outperforms traditional IDS, deep learning anomaly detectors, and single-agent RL baselines across key metrics, achieving a 97.2% detection rate while reducing false positives by over 80% and improving response times by a factor of four. Crucially, it demonstrates superior performance in maintaining physical process stability, presenting a robust new paradigm for securing critical national infrastructure.

Zijie Liu, Xinyu Zhao, Jie Peng et al.

Current medical AI systems often fail to replicate real-world clinical reasoning, as they are predominantly trained and evaluated on static text and question-answer tasks. These tuning methods and benchmarks overlook critical aspects like evidence-based reasoning and handling distracting information. To bridge this gap, we introduce a novel benchmark that simulates real-world diagnostic scenarios, integrating noise and difficulty levels aligned with USMLE standards. Moreover, we explore dialogue-based fine-tuning, which transforms static datasets into conversational formats to better capture iterative reasoning processes. Experiments show that dialogue-tuned models outperform traditional methods, with improvements of $9.64\%$ in multi-round reasoning scenarios and $6.18\%$ in accuracy in a noisy environment. Our findings highlight dialogue tuning as a promising approach for advancing clinically aligned and robust medical AI systems.

Junrui Zhang, Xinyu Zhao, Jie Peng et al.

Multimodal learning has shown significant superiority on various tasks by integrating multiple modalities. However, the interdependencies among modalities increase the susceptibility of multimodal models to adversarial attacks. Existing methods mainly focus on attacks on specific modalities or indiscriminately attack all modalities. In this paper, we find that these approaches ignore the differences between modalities in their contribution to final robustness, resulting in suboptimal robustness performance. To bridge this gap, we introduce Vulnerability-Aware Robust Multimodal Adversarial Training (VARMAT), a probe-in-training adversarial training method that improves multimodal robustness by identifying the vulnerability of each modality. To be specific, VARMAT first explicitly quantifies the vulnerability of each modality, grounded in a first-order approximation of the attack objective (Probe). Then, we propose a targeted regularization term that penalizes modalities with high vulnerability, guiding robust learning while maintaining task accuracy (Training). We demonstrate the enhanced robustness of our method across multiple multimodal datasets involving diverse modalities. Finally, we achieve {12.73%, 22.21%, 11.19%} robustness improvement on three multimodal datasets, revealing a significant blind spot in multimodal adversarial training.

Jiaye Yang, Xinyu Zhao, Tianlong Chen et al.

While Artificial Intelligence (AI) shows promise in healthcare applications, existing conversational systems often falter in complex and sensitive medical domains such as Sexual and Reproductive Health (SRH). These systems frequently struggle with hallucination and lack the specialized knowledge required, particularly for sensitive SRH topics. Furthermore, current AI approaches in healthcare tend to prioritize diagnostic capabilities over comprehensive patient care and education. Addressing these gaps, this work at the UNC School of Nursing introduces SARHAchat, a proof-of-concept Large Language Model (LLM)-based chatbot. SARHAchat is designed as a reliable, user-centered system integrating medical expertise with empathetic communication to enhance SRH care delivery. Our evaluation demonstrates SARHAchat's ability to provide accurate and contextually appropriate contraceptive counseling while maintaining a natural conversational flow. The demo is available at https://sarhachat.com/}{https://sarhachat.com/.

Xu Shen, Qi Zhang, Song Wang et al.

Large Language Model based multi-agent systems (MAS) excel at collaborative problem solving but remain brittle to cascading errors: a single faulty step can propagate across agents and disrupt the trajectory. In this paper, we present MASC, a metacognitive framework that endows MAS with real-time, unsupervised, step-level error detection and self-correction. MASC rethinks detection as history-conditioned anomaly scoring via two complementary designs: (1) Next-Execution Reconstruction, which predicts the embedding of the next step from the query and interaction history to capture causal consistency, and (2) Prototype-Guided Enhancement, which learns a prototype prior over normal-step embeddings and uses it to stabilize reconstruction and anomaly scoring under sparse context (e.g., early steps). When an anomaly step is flagged, MASC triggers a correction agent to revise the acting agent's output before information flows downstream. On the Who&When benchmark, MASC consistently outperforms all baselines, improving step-level error detection by up to 8.47% AUC-ROC ; When plugged into diverse MAS frameworks, it delivers consistent end-to-end gains across architectures, confirming that our metacognitive monitoring and targeted correction can mitigate error propagation with minimal overhead.

Weiqing Luo, Zhen Tan, Yifan Li et al.

Real-world vision-language applications demand varying levels of perceptual granularity. However, most existing visual large language models (VLLMs), such as LLaVA, pre-assume a fixed resolution for downstream tasks, which leads to subpar performance. To address this problem, we first conduct a comprehensive and pioneering investigation into the resolution preferences of different vision-language tasks, revealing a correlation between resolution preferences with image complexity, and uncertainty variance of the VLLM at different image input resolutions. Building on this insight, we propose an empirical formula to determine the optimal resolution for a given vision-language task, combining these two factors. Second, based on rigorous experiments, we propose a novel parameter-efficient fine-tuning technique to extend the visual input resolution of pre-trained VLLMs to the identified optimal resolution. Extensive experiments on various vision-language tasks validate the effectiveness of our method.

Tianxiang Xu, Zhichao Wen, Xinyu Zhao et al.

The escalating complexity of network threats and the inherent class imbalance in traffic data present formidable challenges for modern Intrusion Detection Systems (IDS). While Graph Neural Networks (GNNs) excel in modeling topological structures and Temporal Convolutional Networks (TCNs) are proficient in capturing time-series dependencies, a framework that synergistically integrates both while explicitly addressing data imbalance remains an open challenge. This paper introduces a novel deep learning framework, named Gated Temporal Convolutional Network and Graph (GTCN-G), engineered to overcome these limitations. Our model uniquely fuses a Gated TCN (G-TCN) for extracting hierarchical temporal features from network flows with a Graph Convolutional Network (GCN) designed to learn from the underlying graph structure. The core innovation lies in the integration of a residual learning mechanism, implemented via a Graph Attention Network (GAT). This mechanism preserves original feature information through residual connections, which is critical for mitigating the class imbalance problem and enhancing detection sensitivity for rare malicious activities (minority classes). We conducted extensive experiments on two public benchmark datasets, UNSW-NB15 and ToN-IoT, to validate our approach. The empirical results demonstrate that the proposed GTCN-G model achieves state-of-the-art performance, significantly outperforming existing baseline models in both binary and multi-class classification tasks.

Xu Shen, Song Wang, Zhen Tan et al.

Large language models (LLMs) increasingly rely on Chain-of-Thought (CoT) prompting to improve problem-solving and provide seemingly transparent explanations. However, growing evidence shows that CoT often fail to faithfully represent the underlying reasoning process, raising concerns about their reliability in high-risk applications. Although prior studies have focused on mechanism-level analyses showing that CoTs can be unfaithful, they leave open the practical challenge of deciding whether a specific trajectory is faithful to the internal reasoning of the model. To address this gap, we introduce FaithCoT-Bench, a unified benchmark for instance-level CoT unfaithfulness detection. Our framework establishes a rigorous task formulation that formulates unfaithfulness detection as a discriminative decision problem, and provides FINE-CoT (Faithfulness instance evaluation for Chain-of-Thought), an expert-annotated collection of over 1,000 trajectories generated by four representative LLMs across four domains, including more than 300 unfaithful instances with fine-grained causes and step-level evidence. We further conduct a systematic evaluation of eleven representative detection methods spanning counterfactual, logit-based, and LLM-as-judge paradigms, deriving empirical insights that clarify the strengths and weaknesses of existing approaches and reveal the increased challenges of detection in knowledge-intensive domains and with more advanced models. To the best of our knowledge, FaithCoT-Bench establishes the first comprehensive benchmark for instance-level CoT faithfulness, setting a solid basis for future research toward more interpretable and trustworthy reasoning in LLMs.

Xinyu Zhao, Zhen Tan, Maya Enisman et al.

Successful group meetings, such as those implemented in group behavioral-change programs, work meetings, and other social contexts, must promote individual goal setting and execution while strengthening the social relationships within the group. Consequently, an ideal facilitator must be sensitive to the subtle dynamics of disengagement, difficulties with individual goal setting and execution, and interpersonal difficulties that signal a need for intervention. The challenges and cognitive load experienced by facilitators create a critical gap for an embodied technology that can interpret social exchanges while remaining aware of the needs of the individuals in the group and providing transparent recommendations that go beyond powerful but "black box" foundation models (FMs) that identify social cues. We address this important demand with a social robot co-facilitator that analyzes multimodal meeting data and provides discreet cues to the facilitator. The robot's reasoning is powered by an agentic concept bottleneck model (CBM), which makes decisions based on human-interpretable concepts like participant engagement and sentiments, ensuring transparency and trustworthiness. Our core contribution is a transfer learning framework that distills the broad social understanding of an FM into our specialized and transparent CBM. This concept-driven system significantly outperforms direct zero-shot FMs in predicting the need for intervention and enables real-time human correction of its reasoning. Critically, we demonstrate robust knowledge transfer: the model generalizes across different groups and successfully transfers the expertise of senior human facilitators to improve the performance of novices. By transferring an expert's cognitive model into an interpretable robotic partner, our work provides a powerful blueprint for augmenting human capabilities in complex social domains.

Xinyu Zhao, Jun Liu, Faqiang Wang et al.

The elliptical shape prior information plays a vital role in improving the accuracy of image segmentation for specific tasks in medical and natural images. Existing deep learning-based segmentation methods, including the Segment Anything Model (SAM), often struggle to produce segmentation results with elliptical shapes efficiently. This paper proposes a new approach to integrate the prior of elliptical shapes into the deep learning-based SAM image segmentation techniques using variational methods. The proposed method establishes a parameterized elliptical contour field, which constrains the segmentation results to align with predefined elliptical contours. Utilizing the dual algorithm, the model seamlessly integrates image features with elliptical priors and spatial regularization priors, thereby greatly enhancing segmentation accuracy. By decomposing SAM into four mathematical sub-problems, we integrate the variational ellipse prior to design a new SAM network structure, ensuring that the segmentation output of SAM consists of elliptical regions. Experimental results on some specific image datasets demonstrate an improvement over the original SAM.

Jinhao Duan, Xinyu Zhao, Zhuoxuan Zhang et al.

Although Large Language Models (LLMs) succeed in human-guided conversations such as instruction following and question answering, the potential of LLM-guided conversations-where LLMs direct the discourse and steer the conversation's objectives-remains under-explored. In this study, we first characterize LLM-guided conversation into three fundamental components: (i) Goal Navigation; (ii) Context Management; (iii) Empathetic Engagement, and propose GuideLLM as an installation. We then implement an interviewing environment for the evaluation of LLM-guided conversation. Specifically, various topics are involved in this environment for comprehensive interviewing evaluation, resulting in around 1.4k turns of utterances, 184k tokens, and over 200 events mentioned during the interviewing for each chatbot evaluation. We compare GuideLLM with 6 state-of-the-art LLMs such as GPT-4o and Llama-3-70b-Instruct, from the perspective of interviewing quality, and autobiography generation quality. For automatic evaluation, we derive user proxies from multiple autobiographies and employ LLM-as-a-judge to score LLM behaviors. We further conduct a human-involved experiment by employing 45 human participants to chat with GuideLLM and baselines. We then collect human feedback, preferences, and ratings regarding the qualities of conversation and autobiography. Experimental results indicate that GuideLLM significantly outperforms baseline LLMs in automatic evaluation and achieves consistent leading performances in human ratings.

Song Wang, Zhen Tan, Xinyu Zhao et al.

Graph generation generally aims to create new graphs that closely align with a specific graph distribution. Existing works often implicitly capture this distribution through the optimization of generators, potentially overlooking the intricacies of the distribution itself. Furthermore, these approaches generally neglect the insights offered by the learned distribution for graph generation. In contrast, in this work, we propose a novel self-conditioned graph generation framework designed to explicitly model graph distributions and employ these distributions to guide the generation process. We first perform self-conditioned modeling to capture the graph distributions by transforming each graph sample into a low-dimensional representation and optimizing a representation generator to create new representations reflective of the learned distribution. Subsequently, we leverage these bootstrapped representations as self-conditioned guidance for the generation process, thereby facilitating the generation of graphs that more accurately reflect the learned distributions. We conduct extensive experiments on generic and molecular graph datasets across various fields. Our framework demonstrates superior performance over existing state-of-the-art graph generation methods in terms of graph quality and fidelity to training data.

Xinyu Zhao, Sa Huang, Wei Pang et al.

We study the problem of registration for medical CT images from a novel perspective -- the sensitivity to degree of deformations in CT images. Although some learning-based methods have shown success in terms of average accuracy, their ability to handle regions with local large deformation (LLD) may significantly decrease compared to dealing with regions with minor deformation. This motivates our research into this issue. Two main causes of LLDs are organ motion and changes in tissue structure, with the latter often being a long-term process. In this paper, we propose a novel registration model called Cascade-Dilation Inter-Layer Differential Network (CDIDN), which exhibits both high deformation impedance capability (DIC) and accuracy. CDIDN improves its resilience to LLDs in CT images by enhancing LLDs in the displacement field (DF). It uses a feature-based progressive decomposition of LLDs, blending feature flows of different levels into a main flow in a top-down manner. It leverages Inter-Layer Differential Module (IDM) at each level to locally refine the main flow and globally smooth the feature flow, and also integrates feature velocity fields that can effectively handle feature deformations of various degrees. We assess CDIDN using lungs as representative organs with large deformation. Our findings show that IDM significantly enhances LLDs of the DF, by which improves the DIC and accuracy of the model. Compared with other outstanding learning-based methods, CDIDN exhibits the best DIC and excellent accuracy. Based on vessel enhancement and enhanced LLDs of the DF, we propose a novel method to accurately track the appearance, disappearance, enlargement, and shrinkage of pulmonary lesions, which effectively addresses detection of early lesions and peripheral lung lesions, issues of false enlargement, false shrinkage, and mutilation of lesions.

Qijie Wei, Jingyuan Yang, Bo Wang et al.

This paper addresses the emerging task of recognizing multiple retinal diseases from wide-field (WF) and ultra-wide-field (UWF) fundus images. For an effective use of existing large amount of labeled color fundus photo (CFP) data and the relatively small amount of WF and UWF data, we propose a supervised domain adaptation method named Cross-domain Collaborative Learning (CdCL). Inspired by the success of fixed-ratio based mixup in unsupervised domain adaptation, we re-purpose this strategy for the current task. Due to the intrinsic disparity between the field-of-view of CFP and WF/UWF images, a scale bias naturally exists in a mixup sample that the anatomic structure from a CFP image will be considerably larger than its WF/UWF counterpart. The CdCL method resolves the issue by Scale-bias Correction, which employs Transformers for producing scale-invariant features. As demonstrated by extensive experiments on multiple datasets covering both WF and UWF images, the proposed method compares favorably against a number of competitive baselines.

Xinyu Zhao, Razvan C. Fetecau, Mo Chen

Collaborative autonomous multi-agent systems covering a specified area have many potential applications, such as UAV search and rescue, forest fire fighting, and real-time high-resolution monitoring. Traditional approaches for such coverage problems involve designing a model-based control policy based on sensor data. However, designing model-based controllers is challenging, and the state-of-the-art classical control policy still exhibits a large degree of sub-optimality. In this paper, we present a reinforcement learning (RL) approach for the multi-agent efficient domain coverage problem involving agents with second-order dynamics. Our approach is based on the Multi-Agent Proximal Policy Optimization Algorithm (MAPPO). Our proposed network architecture includes the incorporation of LSTM and self-attention, which allows the trained policy to adapt to a variable number of agents. Our trained policy significantly outperforms the state-of-the-art classical control policy. We demonstrate our proposed method in a variety of simulated experiments.

Xinyu Zhao, Hao Yan, Zhiyong Hu et al.

Electrical conduction among cardiac tissue is commonly modeled with partial differential equations, i.e., reaction-diffusion equation, where the reaction term describes cellular stimulation and diffusion term describes electrical propagation. Detecting and identifying of cardiac cells that produce abnormal electrical impulses in such nonlinear dynamic systems are important for efficient treatment and planning. To model the nonlinear dynamics, simulation has been widely used in both cardiac research and clinical study to investigate cardiac disease mechanisms and develop new treatment designs. However, existing cardiac models have a great level of complexity, and the simulation is often time-consuming. We propose a deep spatio-temporal sparse decomposition (DSTSD) approach to bypass the time-consuming cardiac partial differential equations with the deep spatio-temporal model and detect the time and location of the anomaly (i.e., malfunctioning cardiac cells). This approach is validated from the data set generated from the Courtemanche-Ramirez-Nattel (CRN) model, which is widely used to model the propagation of the transmembrane potential across the cross neuron membrane. The proposed DSTSD achieved the best accuracy in terms of spatio-temporal mean trend prediction and anomaly detection.

Mingliang Bai, Xinyu Zhao, Zhenhua Long et al.

Solar energy is a clean and renewable energy. Photovoltaic (PV) power is an important way to utilize solar energy. Accurate PV power forecast is crucial to the large-scale application of PV power and the stability of electricity grid. This paper proposes a novel method for short-term photovoltaic power forecast using deep convolutional long short-term memory (ConvLSTM) network and kernel density estimation (KDE). In the proposed method, ConvLSTM is used to forecast the future photovoltaic power and KDE is used for estimating the joint probabilistic density function and giving the probabilistic confidence interval. Experiments in an actual photovoltaic power station verify the effectiveness of the proposed method. Comparison experiments with convolutional neural network (CNN) and long short-term memory network (LSTM)shows that ConvLSTM can combine the advantages of both CNN and LSTM and significantly outperform CNN and LSTM in terms of forecast accuracy. Through further comparison with other five conventional methods including multilayer perceptron (MLP), support vector regression (SVR), extreme learning machine (ELM), classification and regression tree (CART) and gradient boosting decision tree (GBDT), ConvLSTM can significantly improve the forecast accuracy by more than 20% for most of the five methods and the superiorities of ConvLSTM are further verified.

Kaj Bostrom, Xinyu Zhao, Swarat Chaudhuri et al.

An interpretable system for open-domain reasoning needs to express its reasoning process in a transparent form. Natural language is an attractive representation for this purpose -- it is both highly expressive and easy for humans to understand. However, manipulating natural language statements in logically consistent ways is hard: models must cope with variation in how meaning is expressed while remaining precise. In this paper, we describe ParaPattern, a method for building models to generate deductive inferences from diverse natural language inputs without direct human supervision. We train BART-based models (Lewis et al., 2020) to generate the result of applying a particular logical operation to one or more premise statements. Crucially, we develop a largely automated pipeline for constructing suitable training examples from Wikipedia. We evaluate our models using out-of-domain sentence compositions from the QASC (Khot et al., 2020) and EntailmentBank (Dalvi et al., 2021) datasets as well as targeted perturbation sets. Our results show that our models are substantially more accurate and flexible than baseline systems. ParaPattern achieves 85% validity on examples of the 'substitution' operation from EntailmentBank without the use of any in-domain training data, matching the performance of a model fine-tuned for EntailmentBank. The full source code for our method is publicly available.

Xinyu Zhao, Shih-ting Lin, Greg Durrett

A principal barrier to training temporal relation extraction models in new domains is the lack of varied, high quality examples and the challenge of collecting more. We present a method of automatically collecting distantly-supervised examples of temporal relations. We scrape and automatically label event pairs where the temporal relations are made explicit in text, then mask out those explicit cues, forcing a model trained on this data to learn other signals. We demonstrate that a pre-trained Transformer model is able to transfer from the weakly labeled examples to human-annotated benchmarks in both zero-shot and few-shot settings, and that the masking scheme is important in improving generalization.