How to train machine learning agents to do fun and absurd things using Unity3D by A. Chatzizacharias

Learn how to create fun and absurd games using machine learning agents, PyTorch, and Unity. Discover PPO algorithm, reinforcement learning, and imitation learning techniques to train your agents and create immersive game experiences.

Key takeaways

PyTorch and Unity can be used together to create games involving machine learning agents.
PPO (Proximal Policy Optimization) is a type of reinforcement learning algorithm that can be used to train agents.
Machine learning agents can be used to create fun and absurd games, such as a game where an agent tries to catch a ball with a sword.
The Unity editor can be used to create a visual representation of the game and train the agent using the Python trainer.
The agent’s instructions can be based on the input received from the game, such as the position of the ball.
The agent can be trained using reinforcement learning, where it is rewarded or punished based on its performance.
The agent can also be trained using imitation learning, where it tries to imitate the behavior of a human player.
The ML-Agents package is an open-source package that allows developers to create and train machine learning agents for games.
The package includes a range of features, such as reinforcement learning, imitation learning, and automated game testing.
The package can also be used to create procedural content, such as generating maps or items within a game.
The package can be used to create fun and immersive game experiences that are not possible with traditional game development techniques.

How to train machine learning agents to do fun and absurd things using Unity3D by A. Chatzizacharias

More talks