How we used Reinforcement Learning to solve the Abbey of Crime | Juantomás Garcia Molina

Learn how to solve complex problems using reinforcement learning and deep learning, as demonstrated by the Abbey of Crime project, where a neural network plays a difficult game with billions of possible moves.

Key takeaways

Using Keras is recommended for deep learning projects.
Reinforcement learning is a method to learn from rewards or punishments.
The Abbey of Crime project uses a neural network to play a game.
The game is very difficult to solve, with billions of possible moves.
The project uses a Docker file to create a container for the game engine.
The game engine is run in a Kubernetes cluster with 10 machines.
The project uses a smart cluster to manage the game engine and artificial intelligence.
The artificial intelligence needs to learn from rewards or punishments to improve its game-playing abilities.
The project uses a OpenAI gene to normalize the game environment.
The game is played in a loop, with the artificial intelligence making decisions and receiving rewards or punishments.
The project uses a fake keyboard to interact with the game engine.
The game engine runs in a cluster with multiple machines.
The project uses a neural network to predict the outcome of the game.
The artificial intelligence needs to learn from millions of games to improve its abilities.

How we used Reinforcement Learning to solve the Abbey of Crime | Juantomás Garcia Molina

More talks