Playing ATARI games with deep reinforcement learning

The aim of this workshop is to expose the audience to state of the art AI and use programming as a powerful tool for teaching advanced AI concepts. In this workshop we will re-implement the first ground breaking paper from DeepMind and train a DRL agent to play ATARI games. Participants in the workshop will not only immerse themselves in one of the hottest topics in AI, but also learn useful concepts such as differentiable programming which they can later use in their own projects. The workshop is meant to be an interactive coding session where we will discuss and implement together the DRL agent step by step. The workshop will be split in 3 parts:

1. Introduction to reinforcement learning and OpenAI Gym
2. Introduction to building deep neural networks with Chainer
3. Implementing a deep reinforcement learning agent that learns to play the ATARI game Pong.

Expected duration of the workshop is ~3 hours with 2 x 15-minute breaks.