Architecting and Visualizing Deep Reinforcement Learning Models

TL;DR

This paper presents the development of a flexible Atari Pong environment, a DRL model, and an interactive visualization tool to enhance understanding of Deep Reinforcement Learning mechanics.

Contribution

It introduces a new Pong environment, addresses data challenges, and creates an interactive visualization for better DRL comprehension.

Findings

Successfully built a DRL Pong agent with visualization

Identified and addressed data deficiencies in new environments

Created an interactive exhibit to demonstrate DRL inference mechanics

Abstract

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.