This repo exists to capture the core ML side of ImprovementML, used in conjunction with the Unity backed game. The architecture I've designed is similar to Unity's ML Agents: a 2 way communication channel is established over TCP sockets, with a basic exchange protocol between the python learner and the unity agent.
See stevenire.land for project updates or technical progress.