Contribute Media
A thank you to everyone who makes this possible: Read More

(Alpha) Zero to Elo

Description

Reinforcement learning is a sequential decision making framework that has received a lot of attention given incredible results from DeepMind, DQN and AlphaGo. More recently DeepMind delivered a more general AI game player call AlphaZero, which learns by self-play.

During this session we will introduce you the key ideas behind AlphaZero:

  • Reinforcement Learning and Value function
  • MonteCarlo Tree Search
  • Self-play

We will explore how they contribute to beat greatest Go player of our time, in Python.

Agenda:

  1. Introduction to AI in games
  2. What AlphaZero is made of? (Boring theory)
  3. How to build your own AlphaZero? (Cool code)

in __on domenica 22 aprile at 15:30 **See schedule**

Details

Improve this page