A reinforcement learning toolkit for automating tasks on desktop applications.
Final year project for Computer Science at UWE.
There are two main components of this project: the desktop client, and the web
services. Located in client
and web
respectively.
The desktop client is a PyGTK application that enables users to create and train their own RL agents based on recorded experience with other desktop applications. The web services provide utilities for saving agent parameter backups and offloading taxing compute to cloud compute runners.