Recently, a sub-forum of the World Smart Congress, the Multiagent Systems and Deep Reinforcement Learning Forum, was held in the Xie Kechang lecture Hall of Tianjin University. Its theme was "Multi-Agent Systems and Deep Reinforcement Learning." 

Experts from universities and institutes at home and abroad, including Tsinghua University, the Chinese University of Hong Kong and Nanyang Technological University, as well as students from universities across the country participated in the forum.  

The forum was chaired by Prof. Hao Jianye of Tianjin University. Prof. Wang Zan, Dean of the Institute of New Media Research, attended the forum and delivered a speech. Nine participating experts gave keynote speeches.  

Participants listening to the lecture 

During the Forum, Professor Gao Yang from Nanjing University first analyzed the importance of the multiagent system in AI ？？intelligence systems.  

Then, Prof. An Bo from Nanyang Technological University introduced large-scale game calculation to the audience. He discussed challenges that large-scale game calculation faces, major progress that has been made in algorithm design in recent years, and major successes in Texas Peak and security. Finally, he provided an outlook of the future of large-scale game research and its potential applications.  

After a brief coffee break, Wang Wenfeng, director of the China Electronics Standardization Institute gave a report titled "Standardization of Biometrics Technology" and shared related standardization work. Prof. Zhang Chongjie of Tsinghua University provided another report entitled "Towards Task Transfer and Generalization in Deep Reinforcement Learning", discussing the implementation of task migration in model-based and model-free RL. Researcher Qin Tao of Microsoft Research Asia gave a speech titled "Deep Reinforcement Learning: Challenges and Opportunities", discussing in depth the challenges and opportunities currently faced in deep reinforcement learning. 

Prof. Liang Haofeng of the Chinese University of Hong Kong gives a report 

In the afternoon, Prof. Liang Haofeng of the Chinese University of Hong Kong gave a report titled "Norm Non-Emergence as a Stable State in Multiagent Systems", through which he shared the solution to problems like contracts in the multiagent system.  

Researcher Zhao Dongbin of the Institute of Automation at the Chinese Academy of Sciences focused on the topic of “Deep Reinforcement Learning Progress: From AlphaGo to AlphaGo Zero”, under which he introduced the research progress of deep reinforcement learning from AlphaGo to AlphaGo Zero, and showed applications of deep reinforcement learning in games, smart driving, robots, and medical problems.  

Professor Tang Pingzhong of Tsinghua University introduced the design of large-scale application mechanisms. Finally, Professor Yu Yang of Nanjing University spoke about how reinforcement learning goes to reality, and analyzed the research progress of migration from virtual environments to the physical environment and virtualization of physical environments. 

Experts and participants discuss topics 

During the discussion, participating experts and other participants discussed future research challenges the multiagent system could face. Some students asked questions about scientific research, and the experts answered correspondingly.  

The forum was sponsored by the Tianjin Municipal Education Commission and co-organized by the Tianjin Haihe Education Park Management Committee and Tianjin University.  

By: Zhao Han

Editors: Qin Mian and Keith Harrington