Developing a generalized intelligent agent by processing information on webpages

Abstract

In this thesis, I designed and implemented a framework for reinforcement learning (RL) agents to interact with a web environment. With this framework, I introduce a new challenge for RL agents to learn human activity on the web. By defining a series of tasks such as using the web as a navigable resource to perform actions, I introduce an extension to natural language processing models. The framework provides an agent with rich features including element positioning, color, and size in order to process text represented in a 2D web space..