Project 26: Games and Control in Uncertain Environments

Contact Information:

Prof. Jianping He 

Email: jphe@sjtu.edu.cn

Project Description and Objectives:

This summer research program for undergraduates delves into decision-making under uncertainty using Markov Decision Processes (MDPs) and Reinforcement Learning (RL). Students will explore the theoretical foundations of MDPs and apply them to real-world scenarios where outcomes are probabilistic. They will also engage with RL algorithms to train agents to make optimal decisions in dynamic environments with unknown parameters. The project aims to bridge the gap between theoretical knowledge and practical applications, equipping students with skills to tackle complex decision-making problems in AI.

Eligibility Requirements:

The project requires students to have programming skills in languages such as Python, as well as a mathematical foundation in reinforcement learning, stochastic processes, and game theory. Additionally, a basic understanding of control theory is also required.

Main Tasks:

Analyze and understand the theoretical underpinnings of Markov Decision Processes to model probabilistic outcomes in decision-making scenarios.

Implement and experiment with Reinforcement Learning algorithms to train agents for optimal decision-making in uncertain environments.

Apply theoretical knowledge to practical problems, developing solutions that demonstrate the intersection of AI and complex decision-making.

Website:

Lab: http://iwin-fins.com

School: http://www.seiee.sjtu.edu.cn/