Xiangyu Kong
Embedded Files
I am currently a researcher at Beijing Institute for General Artificial Intelligence (BIGAI). Before that, I worked at Microsoft Research Asia (MSRA) as a researcher in media computing group. I obtained my Ph.D. degree in computer applied technology from Peking University at 2019 under the supervision of Prof. Yizhou Wang, and bachelor's degree in computer science at 2013 from Harbin Institute of Technology, where I was advised by Prof. Weigang Zhang.
Currently, my research focuses on embodied audio/visual perception, multi-modal LLM, LLM agents, multi-agent reinforcement learning.
Richelieu: Self-Evolving LLM-Based Agents for AI Diplomacy
Zhenyu Guan, Xiangyu Kong*, Fangwei Zhong*, Yizhou Wang
NeurIPS, 2024
CivRealm: A Learning and Reasoning Odyssey in Civilization for Decision-Making Agents
Siyuan Qi*, Shuo Chen*, Yexin Li*, Xiangyu Kong*, Junqi Wang*, Bangcheng Yang, Pring Wong, Yifan Zhong, Xiaoyuan Zhang, Zhaowei Zhang, Nian Liu, Yaodong Yang, Song-Chun Zhu
ICLR, 2024 spotlight
DasFormer: Deep Alternating Spectrogram Transformer for Multi/Single-Channel Speech Separation
Shuo Wang, Xiangyu Kong*, Xiulian Peng, Hesam Movassagh, Vinod Prakash and Yan Lu
ICASSP, 2023
Multi-Modal Multi-Correlation Learning for Audio-Visual Speech Separation
Xiaoyu Wang, Xiangyu Kong, Xiulian Peng and Yan Lu
INTERSPEECH, 2022
Pose-Assisted Multi-Camera Collaboration for Active Object Tracking
Jing Li, Jing Xu, Fangwei Zhong, Xiangyu Kong, Yu Qiao and Yizhou Wang
AAAI Conference on Artificial Intelligence (AAAI), 2020
Effective Master-Slave Communication On A Multi-Agent Deep Reinforcement Learning System
Xiangyu Kong, Bo Xin, Fangchen Liu and Yizhou Wang
Neural Information Processing Systems Workshop on Hierarchical Reinforcement Learning (NIPSW), contributed talk, 2017
We revisit the canonical idea of the master-slave architecture for deep multi-agent communication and outperform state-of-the-arts methods on several challenging multi-agent tasks, including several micro-battle tasks on the famous RTS game - StarCraft.
Collaborative Deep Reinforcement Learning for Joint Object Search
Xiangyu Kong, Bo Xin, Yizhou Wang and Gang Hua
Computer Vision and Pattern Recognition (CVPR), 2017
A novel collaborative multi-agent deep reinforcement learning algorithm is proposed to learn the optimal policy for joint active object localization.
Robust Complex Behaviour Modeling at 90Hz
Xiangyu Kong, Yizhou Wang and Tao Xiang
AAAI Conference on Artificial Intelligence (AAAI), Oral Presentation, 2016
We propose a novel approach to complex behaviour modeling based on simple features, joint and iterative spatial, temporal and correlation context modeling, and a set of extremely simple Bernoulli-distribution-based behaviour models.
Page updated
Google Sites
Report abuse