Suhang Wang
YOU?
Author Swipe
View article: Unlocking the Power of Multi-Agent LLM for Reasoning: From Lazy Agents to Deliberation
Unlocking the Power of Multi-Agent LLM for Reasoning: From Lazy Agents to Deliberation Open
Large Language Models (LLMs) trained with reinforcement learning and verifiable rewards have achieved strong results on complex reasoning tasks. Recent work extends this paradigm to a multi-agent setting, where a meta-thinking agent propos…
View article: Generalizing Test-time Compute-optimal Scaling as an Optimizable Graph
Generalizing Test-time Compute-optimal Scaling as an Optimizable Graph Open
Test-Time Scaling (TTS) improves large language models (LLMs) by allocating additional computation during inference, typically through parallel, sequential, or hybrid scaling. However, prior studies often assume fixed collaboration archite…
View article: xTime: Extreme Event Prediction with Hierarchical Knowledge Distillation and Expert Fusion
xTime: Extreme Event Prediction with Hierarchical Knowledge Distillation and Expert Fusion Open
Extreme events frequently occur in real-world time series and often carry significant practical implications. In domains such as climate and healthcare, these events, such as floods, heatwaves, or acute medical episodes, can lead to seriou…
View article: Seeing but Not Believing: Probing the Disconnect Between Visual Attention and Answer Correctness in VLMs
Seeing but Not Believing: Probing the Disconnect Between Visual Attention and Answer Correctness in VLMs Open
Vision-Language Models (VLMs) achieve strong results on multimodal tasks such as visual question answering, yet they can still fail even when the correct visual evidence is present. In this work, we systematically investigate whether these…
View article: TRAJECT-Bench:A Trajectory-Aware Benchmark for Evaluating Agentic Tool Use
TRAJECT-Bench:A Trajectory-Aware Benchmark for Evaluating Agentic Tool Use Open
Large language model (LLM)-based agents increasingly rely on tool use to complete real-world tasks. While existing works evaluate the LLMs' tool use capability, they largely focus on the final answers yet overlook the detailed tool usage t…
View article: SpecDetect: Simple, Fast, and Training-Free Detection of LLM-Generated Text via Spectral Analysis
SpecDetect: Simple, Fast, and Training-Free Detection of LLM-Generated Text via Spectral Analysis Open
The proliferation of high-quality text from Large Language Models (LLMs) demands reliable and efficient detection methods. While existing training-free approaches show promise, they often rely on surface-level statistics and overlook funda…
View article: Generalize across Homophily and Heterophily: Hybrid Spectral Graph Pre-Training and Prompt Tuning
Generalize across Homophily and Heterophily: Hybrid Spectral Graph Pre-Training and Prompt Tuning Open
Graph ``pre-training and prompt-tuning'' aligns downstream tasks with pre-trained objectives to enable efficient knowledge transfer under limited supervision. However, existing methods rely on homophily-based low-frequency knowledge, faili…
View article: A Survey on Small Language Models in the Era of Large Language Models: Architecture, Capabilities, and Trustworthiness
A Survey on Small Language Models in the Era of Large Language Models: Architecture, Capabilities, and Trustworthiness Open
View article: Are You Using Reliable Graph Prompts? Trojan Prompt Attacks on Graph Neural Networks
Are You Using Reliable Graph Prompts? Trojan Prompt Attacks on Graph Neural Networks Open
View article: Bradley-Terry and Multi-Objective Reward Modeling Are Complementary
Bradley-Terry and Multi-Objective Reward Modeling Are Complementary Open
Reward models trained on human preference data have demonstrated strong effectiveness in aligning Large Language Models (LLMs) with human intent under the framework of Reinforcement Learning from Human Feedback (RLHF). However, RLHF remain…
View article: Image Corruption-Inspired Membership Inference Attacks against Large Vision-Language Models
Image Corruption-Inspired Membership Inference Attacks against Large Vision-Language Models Open
Large vision-language models (LVLMs) have demonstrated outstanding performance in many downstream tasks. However, LVLMs are trained on large-scale datasets, which can pose privacy risks if training images contain sensitive information. The…
View article: SUA: Stealthy Multimodal Large Language Model Unlearning Attack
SUA: Stealthy Multimodal Large Language Model Unlearning Attack Open
Multimodal Large Language Models (MLLMs) trained on massive data may memorize sensitive personal information and photos, posing serious privacy risks. To mitigate this, MLLM unlearning methods are proposed, which fine-tune MLLMs to reduce …
View article: BioMol-MQA: A Multi-Modal Question Answering Dataset For LLM Reasoning Over Bio-Molecular Interactions
BioMol-MQA: A Multi-Modal Question Answering Dataset For LLM Reasoning Over Bio-Molecular Interactions Open
Retrieval augmented generation (RAG) has shown great power in improving Large Language Models (LLMs). However, most existing RAG-based LLMs are dedicated to retrieving single modality information, mainly text; while for many real-world pro…
View article: Attention Knows Whom to Trust: Attention-based Trust Management for LLM Multi-Agent Systems
Attention Knows Whom to Trust: Attention-based Trust Management for LLM Multi-Agent Systems Open
Large Language Model-based Multi-Agent Systems (LLM-MAS) have demonstrated strong capabilities in solving complex tasks but remain vulnerable when agents receive unreliable messages. This vulnerability stems from a fundamental gap: LLM age…
View article: Unlearning Inversion Attacks for Graph Neural Networks
Unlearning Inversion Attacks for Graph Neural Networks Open
Graph unlearning methods aim to efficiently remove the impact of sensitive data from trained GNNs without full retraining, assuming that deleted information cannot be recovered. In this work, we challenge this assumption by introducing the…
View article: Keeping an Eye on LLM Unlearning: The Hidden Risk and Remedy
Keeping an Eye on LLM Unlearning: The Hidden Risk and Remedy Open
Although Large Language Models (LLMs) have demonstrated impressive capabilities across a wide range of tasks, growing concerns have emerged over the misuse of sensitive, copyrighted, or harmful data during training. To address these concer…
View article: GPR: Empowering Generation with Graph-Pretrained Retriever
GPR: Empowering Generation with Graph-Pretrained Retriever Open
Graph retrieval-augmented generation (GRAG) places high demands on graph-specific retrievers. However, existing retrievers often rely on language models pretrained on plain text, limiting their effectiveness due to domain misalignment and …
View article: Bridging Source and Target Domains via Link Prediction for Unsupervised Domain Adaptation on Graphs
Bridging Source and Target Domains via Link Prediction for Unsupervised Domain Adaptation on Graphs Open
Graph neural networks (GNNs) have shown great ability for node classification on graphs. However, the success of GNNs relies on abundant labeled data, while obtaining high-quality labels is costly and challenging, especially for newly emer…
View article: Diagnosing and Addressing Pitfalls in KG-RAG Datasets: Toward More Reliable Benchmarking
Diagnosing and Addressing Pitfalls in KG-RAG Datasets: Toward More Reliable Benchmarking Open
Knowledge Graph Question Answering (KGQA) systems rely on high-quality benchmarks to evaluate complex multi-hop reasoning. However, despite their widespread use, popular datasets such as WebQSP and CWQ suffer from critical quality issues, …
View article: Fairness-aware Prompt Tuning for Graph Neural Networks
Fairness-aware Prompt Tuning for Graph Neural Networks Open
View article: Towards Graph Foundation Models: A Transferability Perspective
Towards Graph Foundation Models: A Transferability Perspective Open
In recent years, Graph Foundation Models (GFMs) have gained significant attention for their potential to generalize across diverse graph domains and tasks. Some works focus on Domain-Specific GFMs, which are designed to address a variety o…
View article: A General Framework to Enhance Fine-tuning-based LLM Unlearning
A General Framework to Enhance Fine-tuning-based LLM Unlearning Open
Unlearning has been proposed to remove copyrighted and privacy-sensitive data from Large Language Models (LLMs). Existing approaches primarily rely on fine-tuning-based methods, which can be categorized into gradient ascent-based (GA-based…
View article: How Far are LLMs from Real Search? A Comprehensive Study on Efficiency, Completeness, and Inherent Capabilities
How Far are LLMs from Real Search? A Comprehensive Study on Efficiency, Completeness, and Inherent Capabilities Open
Search plays a fundamental role in problem-solving across various domains, with most real-world decision-making problems being solvable through systematic search. Drawing inspiration from recent discussions on search and learning, we syste…
View article: Stepwise Perplexity-Guided Refinement for Efficient Chain-of-Thought Reasoning in Large Language Models
Stepwise Perplexity-Guided Refinement for Efficient Chain-of-Thought Reasoning in Large Language Models Open
Chain-of-Thought (CoT) reasoning, which breaks down complex tasks into intermediate reasoning steps, has significantly enhanced the performance of large language models (LLMs) on challenging tasks. However, the detailed reasoning process i…
View article: LanP: Rethinking the Impact of Language Priors in Large Vision-Language Models
LanP: Rethinking the Impact of Language Priors in Large Vision-Language Models Open
Large Vision-Language Models (LVLMs) have shown impressive performance in various tasks. However, LVLMs suffer from hallucination, which hinders their adoption in the real world. Existing studies emphasized that the strong language priors …
View article: Graph-based Molecular In-context Learning Grounded on Morgan Fingerprints
Graph-based Molecular In-context Learning Grounded on Morgan Fingerprints Open
In-context learning (ICL) effectively conditions large language models (LLMs) for molecular tasks, such as property prediction and molecule captioning, by embedding carefully selected demonstration examples into the input prompt. This appr…
View article: Counterfactual Learning on Graphs: A Survey
Counterfactual Learning on Graphs: A Survey Open
Graph-structured data are pervasive in the real-world such as social networks, molecular graphs and transaction networks. Graph neural networks (GNNs) have achieved great success in representation learning on graphs, facilitating various d…
View article: GAMIC: Graph-Aligned Molecular In-context Learning for Molecule Analysis via LLMs
GAMIC: Graph-Aligned Molecular In-context Learning for Molecule Analysis via LLMs Open
View article: Divide-Verify-Refine: Can LLMs Self-align with Complex Instructions?
Divide-Verify-Refine: Can LLMs Self-align with Complex Instructions? Open
View article: Learning with Less: Knowledge Distillation from Large Language Models via Unlabeled Data
Learning with Less: Knowledge Distillation from Large Language Models via Unlabeled Data Open