đź“„ Publications
My earlier work focused on graph learning and foundation models for structured data, with an emphasis on transfer, robustness, and privacy. My current research centers on large language models, including LLM agents, multi-agent interaction, agentic RL, reasoning, grounding, and evaluation, preference modeling for alignment, and embodied agents such as humanoid VLA systems.
Agentic RL, Multi-Agent Systems, and Agent Benchmarks
This line of work studies agents as complete systems: how they interact in social environments, how they improve through agentic RL, and how they are grounded and evaluated in realistic long-horizon tasks.
Multi-Agent Social Simulation for Proactive Policy Optimization
PolicySim: An LLM-Based Agent Social Simulation Sandbox for Proactive Policy Optimization
Renhong Huang, Ning Tang, Jiarong Xu, Yuxuan Cao, Qingqian Tu, Sheng Guo, Bo Zheng, Huiyuan Liu, Yang Yang.
Oral presentation. In Proceedings of the ACM Web Conference 2026 (WWW’26). [arXiv] [PDF]
Agentic RL for Stronger Exploration in LLM Agents
RAPO: Expanding Exploration for LLM Agents via Retrieval-Augmented Policy Optimization
Siwei Zhang, Yun Xiong, Xi Chen, Zi’an Jia, Renhong Huang, Jiarong Xu, Jiawei Zhang.
Oral presentation. In Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD’26).
Environment-Grounded LLM Agents via Automated Configuration
RAT: RunAnyThing via Fully Automated Environment Configuration
Renhong Huang, Dongdong Hua, Yifei Sun, Sitao Ding, Hanyang Yuan, Daixin Wang, Yang Yang.
arXiv preprint. [arXiv]
Agent Benchmarking for Long-Horizon Strategic Decision Making
PTCG-Bench: Can LLM Agents Master Pokémon Trading Card Game?
Dongdong Hua, Yifei Sun, Renhong Huang, Feng Gao, Chunping Wang, Yang Yang.
arXiv preprint. [arXiv]
Preference Learning and Alignment
Beyond acting well, agents also need to be aligned with nuanced and diverse preferences. This work studies how preference structures can be explicitly organized rather than treated as a flat signal.
Structured Preference Modeling for Diversified Recommendation
Tree of Preferences for Diversified Recommendation
Hanyang Yuan, Ning Tang, Tongya Zheng, Jiarong Xu, Xintong Hu, Renhong Huang, Shunyu Liu, Jiacong Hu, Jiawei Chen, Mingli Song.
In Advances in Neural Information Processing Systems 38 (NeurIPS’25).
Selected Graph Learning Foundations
My earlier work in graph machine learning focused on transfer, data-centric learning, and privacy/safety. I keep a concise selection here.
Graph Domain Adaptation from a Data-Centric Perspective
Can Modifying Data Address Graph Domain Adaptation?
Renhong Huang, Jiarong Xu, Xin Jiang, Ruichuan An, Yang Yang.
Oral presentation. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD’24).
Graph Pre-training from a Data-Centric Perspective
Better with Less: A Data-Centric Perspective on Pre-Training Graph Neural Networks
Jiarong Xu, Renhong Huang, Xin Jiang, Yuxuan Cao, Carl Yang, Chunping Wang, Yang Yang.
In Advances in Neural Information Processing Systems 36 (NeurIPS’23).
Graph Extraction Attack
Extracting Training Data from Molecular Pre-trained Models
Renhong Huang, Jiarong Xu, Zhiming Yang, Xiang Si, Xin Jiang, Hanyang Yuan, Chunping Wang, Yang Yang.
In Advances in Neural Information Processing Systems 37 (NeurIPS’24).