Reinforcement Learning

2024/2025

Учебный год

ENG

Обучение ведется на английском языке

Статус:

Курс по выбору

Когда читается:

2-й курс, 1, 2 модуль

Course Syllabus

Abstract

Reinforcement learning is a vanguard method of machine learning aimed at dynamical applications, ranging from video games to autonomous cars, robots, drones etc.Composed of a so-called agent and environment, it is meant to resemble, in a sense, the behavior of living beings.Agents interact with the environment and optimize their actions to improve rewards.Imagine a video game speed run.An agent, the protagonist, interacts with the game environment and wants to beat the game as fast as possible, by dynamically adjusting his or her controls while learning on-the-fly.Reinforcement learning is truly an interdisciplinary subject that can be studied from different perspectives -- machine learning, control theory, dynamical system theory, pure math etc.In this course, we dive into reinforcement learning for studyng the key principles and implementing them on real examples.Topics covered include policy gradient methods, actor-critic, deep, predictive, safe and other kinds of reinforcement learning.Convergence, safety and stability of reinforcement learning are studied.