The tutorial will demonstrate the role games can play in driving the development of Reinforcement Learning algorithms. The presenter will use the Unity Engine with the ML-Agents toolkit (https://github.com/Unity-Technologies/ml-agents) as an example of how dynamic 3D game environments can be utilized for Machine Learning research. The tutorial will both walk through the motivation for the design of the platform as well as describe how to utilize the platform to perform research not otherwise possible, with a particular focus on scenarios applicable to game design. The tutorial will also include an explanation of the kinds of algorithms being used to train machine learning agents, including various Reinforcement Learning and Supervised Learning methods. All of this will take place in the context of a hands-on creation process, where participants will follow along to design an environment, train an agent within in using ML-Agents, and finally deploy the trained model into a full game scenario which can be played. Attendees will walk away with knowledge of how to conduct Machine Learning research within custom-made 3D game environments, with a focus on Deep Reinforcement Learning methodology which can be applied to solving a variety of tasks in these environments.
Miguel Suau de Castro (Unity Technologies)
Links to websites: