多智能体合作估计与学习

The main teaching content of this course is the theoretical foundation, main algorithms, and application cases of multi-agent cooperative estimation and cooperative reinforcement learning. It is an interdisciplinary course that adapts to the rapid development of the new generation of artificial intelligence, and combines control science and engineering with computer science and engineering. The teaching objective of this course is to enable students to understand the theoretical foundations and algorithmic frameworks of distributed estimation, filtering, online optimization, and reinforcement learning, expand their methods of handling uncertainty in distributed processing systems (especially networked systems), and lay a solid theoretical foundation for quickly entering cutting-edge topics.Multi agent cooperative estimation and learning mainly includes the following parts: 1) Reviewing traditional least squares/minimum variance estimation methods, Kalman filtering methods, and online optimization methods from a learning perspective; 2) Exploring the algorithmic foundations of reinforcement learning from the perspective of estimation theory; 3) Least squares/minimum variance estimation methods, Kalman filtering methods The development of online optimization methods and reinforcement learning methods in distributed frameworks. This course is different from existing control courses such as system identification and optimization methods, as well as machine learning courses. It discusses the above methods in a distributed framework and is a new graduate course content.