Architecting and Visualizing Deep Reinforcement Learning Models
This work provides an educational tool for building intuition about DRL, but it is incremental as it applies existing methods to a new interactive setup.
The authors tackled the challenge of creating an interactive Deep Reinforcement Learning (DRL) exhibit for Atari Pong by developing a new game environment and addressing data deficiencies, resulting in a real-time visualization tool to enhance understanding of DRL mechanics.
To meet the growing interest in Deep Reinforcement Learning (DRL), we sought to construct a DRL-driven Atari Pong agent and accompanying visualization tool. Existing approaches do not support the flexibility required to create an interactive exhibit with easily-configurable physics and a human-controlled player. Therefore, we constructed a new Pong game environment, discovered and addressed a number of unique data deficiencies that arise when applying DRL to a new environment, architected and tuned a policy gradient based DRL model, developed a real-time network visualization, and combined these elements into an interactive display to help build intuition and awareness of the mechanics of DRL inference.