AI Game Playing:Using Deep Reinforcement Learning

---

Abstract

In the evolving landscape of artificial intelligence, Deep Reinforcement Learning (DRL) has carved a niche as a transformative methodology for training agents in complex and dynamic tasks. This study ventures into the realm of video game AI, specifically targeting the renowned and challenging Mario game series. Our research is anchored on deploying and refining cutting-edge DRL techniques, including Deep Q-Networks (DQNs), Advantage Actor-Critic (A3C), and Proximal Policy Optimization (PPO). The primary objective is to cultivate an AI agent with the acumen to adeptly traverse and excel in various levels of Mario games, with the aspiration of achieving and potentially surpassing human-level performance. By leveraging the multifaceted nature and popularity of Mario games, this research contributes to the understanding and advancement of DRL in navigating environments that closely resemble real-world complexities and decision-making scenarios.

---

Keywords

Advantage Actor-Critic AI Game Playing Deep Reinforcement Learning, Deep Q-Networks, Proximal Policy Optimization