上海交通大学学报

上海交通大学学报 ›› 2021, Vol. 55 ›› Issue (Sup.1): 3-4.doi: 10.16183/j.cnki.jsjtu.2021.S1.019

所属专题： 上海交通大学建校125周年科学问题专辑

• 人工智能 • 上一篇    下一篇

智能体自我博弈学习是否存在性能极限?

李少远，殷翔   

  1. 上海交通大学 电子信息与电气工程学院，上海200240
  • 收稿日期:2021-02-04 出版日期:2021-03-28 发布日期:2021-04-09
  • 通讯作者: 殷翔，男，副教授，博士生导师.E-mail：yinxiang@sjtu.edu.cn.
  • 作者简介:李少远（1965-），男，河北省衡水市人，教授，从事智能优化控制研究.
  • 基金资助:
    科技部重大专项（2018AAA0101700），国家自然科学基金（61833012）

Do Agent Self-Game and Learning Have Limitation of Performance?

LI Shaoyuan,YIN Xiang   

  1. School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
  • Received:2021-02-04 Online:2021-03-28 Published:2021-04-09

PDF (PC)

295

摘要： 自我博弈学习是智能涌现的基本方法之一.介绍智能体自我博弈学习的基本思想，回顾该方向的最新研究成果，提出智能体自我博弈学习是否存在极限这一核心科学问题，指出需要融合信息论、控制论等多学科方法，从信息和计算的视角探究智能涌现的根源.

关键词: 人工智能, 博弈, 学习

Abstract: Self-game is one of the most fundamental ways for intelligence emergence. A basic idea of the agent self-game and learning was introduced, and the latest research progress in this direction was expounded. In addition, a key question that whether agent self-game and learning have limitation was proposed. However, it is pointed out that cybernetic and the information theory should be combined to investigate the emergence of intelligence from information and computation perspectives.

Key words: artificial intelligence, game, learning

中图分类号: