Furong Huang

Associate Professor@University of Maryland·美国·OpenReview

研究方向

AI alignment · generative ai · reinforcement learning · deep learning theory · spectral methods · tensor decomposition · mixture models · latent variable models

Furong Huang

MomaGraph: State-Aware Unified Scene Graphs with Vision-Language Models for Embodied Task Planning

Zebra-CoT: A Dataset for Interleaved Vision-Language Reasoning

ROVER: Benchmarking Reciprocal Cross-Modal Reasoning for Omnimodal Generation

PropensityBench: Evaluating Latent Safety Risks in Large Language Models via an Agentic Approach

TrustGen: A Platform of Dynamic Benchmarking on the Trustworthiness of Generative Foundation Models

LLaVA-Critic-R1: Your Critic Model is Secretly a Strong Policy Model

Cut the Overcredit: Precision First Process Rewards for Reasoning LLMs

PersonaLedger: Generating Realistic Financial Transactions with Persona Conditioned LLMs and Rule Grounded Feedback

Hold Onto That Thought: Assessing KV Cache Compression On Reasoning

TimeSqueeze: Dynamic Patching for Efficient Time Series Forecasting

Teach a Reward Model to Correct Itself: Reward Guided Adversarial Failure Discovery for Robust Reward Modeling

Lemon: A Unified and Scalable 3D Multimodal Model for Universal Spatial Understanding

Ladders of Thought: A Self-Evolving Curriculum of Progressively Simplified Reasoning Traces

MAFE: Enabling Equitable Algorithm Design in Multi-Agent Multi-Stage Decision-Making Systems

Advancing Regulation in Artificial Intelligence: An Auction-Based Approach

AegisLLM: Scaling Agentic Systems for Self-Reflective Defense in LLM Security

Mitigating Reward Hacking in Inference-Time Alignment of T2I Diffusion Models via Distributional Regularization

SafeThink: A Key to Safety in Multi-Modal Large Reasoning Models

MORSE: A Suite of Programmatically Controllable Multimodal Reasoning Environments with Steerable Difficulty

Practical Memorization Tests for Detecting Copyrighted Data in Large Language Models

Memory Makes The Poison: Over Memorization Drives Visual Poisoning in LVLMs

Does Thinking More Always Help? Mirage of Test-Time Scaling in Reasoning Models

TraceVLA: Visual Trace Prompting Enhances Spatial-Temporal Awareness for Generalist Robotic Policies

SoTA with Less: MCTS-Guided Sample Selection for Data-Efficient Visual Reasoning Self-Improvement

ViCrit: A Verifiable Reinforcement Learning Proxy Task for Visual Perception in VLMs

GenARM: Reward Guided Generation with Autoregressive Reward Model for Test-Time Alignment

Collab: Controlled Decoding using Mixture of Agents for LLM Alignment

SAIL: Self-improving Efficient Online Alignment of Large Language Models

EnsemW2S: Can an Ensemble of LLMs be Leveraged to Obtain a Stronger LLM?

Benchmarking Vision Language Model Unlearning via Fictitious Facial Identity Dataset

Towards Realistic Mechanisms That Incentivize Federated Participation and Contribution

Auction-Based Regulation for Artificial Intelligence

LIAR: Leveraging Inverse Alignment to Jailbreak LLMs in Seconds

AdvBDGen: Adversarially Fortified Prompt-Specific Fuzzy Backdoor Generator Against LLM Alignment

LSH Tells You What To Discard: An Adaptive Locality-Sensitive Strategy for KV Cache Compression