Reinforcement Learning and Geometric Algorithms
CMPS 499/CSCE 572
General Information
Instructor
Miao Jin Office hours: TR 10:45am - 12:00pm
Times
TR 9:30am - 10:45am
Objectives:
Reinforcement Learning: this class will provide a solid introduction to the field of reinforcement learning (RL), the forefront of artificial intelligence. Students will learn the core challenges and approaches through a combination of lectures, and written and coding assignments.
Specifically: Understand how to formalize a decision optimization task as a RL problem. Learn a variety of RL algorithms. Build RL system for sequential decision making. Understand how to apply neural networks to RL to solve large size decision optimization problems.
Topics include:
- Markov decision processes
- Dynamic programming
- Monte Carlo learning
- Temporal difference learning
- Value function approximation
- Policy gradient
- Integration of learning and planning
Geometric Algorithms: students will study fundamental data structures and algorithms from computational geometry and their applications to problems that occur in practice.
Depending on the remaining lecture hours, topics may include a few of the following:
- Convex hulls
- Line segment intersection
- Triangulating a polygon
- Enclosing circle
- Voronoi diagrams
- Delaunay Triangulations
- Robot Motion Planning
Prerequisites
CMPS 341 Formal Foundations of Computer Science.
Textbooks
An Introduction to Reinforcement Learning, Second edition, in progress (Available free online! ) by Richard S. Sutton and Andrew G. Barto (optional).
Computational Geometry, Algorithms and Applications by Mark de Berg et al
(optional).
Grading
- 50% Reinforcement Learning Project I
- 25% Reinforcement Learning Project II
- 25% Homeworks