Building learning environments for real-world RL
This recipe will teach you how to set up and build WebGym, a World of Bits (WoB)-based OpenAI Gym compatible learning platform for training RL Agents for world wide web-based real-world tasks. WoB is an open domain platform for web-based Agents. For more information about WoB, check out the following link: http://proceedings.mlr.press/v70/shi17a/shi17a.pdf.
WebGym provides learning environments for Agents to perceive the world wide web how we (humans) perceive it – using the pixels rendered on our display screen. The Agent interacts with the environment using keyboard and mouse events as actions. This allows the Agent to experience the world wide web how we do, which means we don't need to make any additional modifications for the Agents to train. This allows us to train RL Agents that can directly work with web-based pages and applications to complete real-world tasks.