Motivation:

Reinforcement Learning (RL) is getting relevant in the field of Machine Learning, playing a fundamental role in a wide range of areas such as autonomous driving, robotics or health-care. Classical RL techniques combined with Deep Learning allow complex systems to perform highly sophisticated tasks, unthinkable only one decade ago.
This course will provide a broad view of the most important state-of-the-art methods and the core challenges that RL is facing nowadays. The goal of this course is to assimilate the key ideas and understand upcoming contributions to the field.

The lectures are distributed in three main blocks: introduction, basic methods and advanced methods. Additionally, emphasis is put on the sensitive subject of evaluating of RL methods. In the final lecture, one of the latest major keystones, AlphaStar, is analyzed.

Introduction

• Introduction to Reinforcement Learning (RL)
• Deep Learning for RL

Basic methods

• Deep Q-Network based methods
• Policy Gradient based methods
• RL and Control as Probabilistic Inference
• Planning with Monte Carlo Tree Search

Advanced methods 

• Exploration in RL
• Imitation and off-policy learning
• Hierarchical RL
• Return decomposition for delayed rewards

Evaluation in RL
A case study: Alphastar