Space Invaders Game Agent

[ayush-09]

Project Request

The Space Invaders game requires the player to control a spaceship and shoot down waves of alien invaders while avoiding their attacks. The DQN agent needs to learn an optimal policy to maximize its score by making strategic decisions on shooting, moving, and avoiding enemy projectiles.

---

Field	Description
About	Reinforcement Learning for Space Invaders Game
Github	ayush-09
Email	ayushvarshney43@gmail.com
Label	Project Request

---

Define You

• [x] GSSOC Participant
• [x] Contributor

Deep Q-Network (DQN) Agent for Space Invaders Game

Description

The goal of this project is to develop a Deep Q-Network (DQN) agent using the OpenAI Gym environment and Keras-RL2 library to play the Space Invaders game. The DQN agent will learn to navigate the game environment and achieve high scores by training on a combination of exploration and exploitation strategies.

Scope

• Set up the Space Invaders game environment using the Gym library.
• Implement the DQN model architecture using Keras, including convolutional and fully connected layers.
• Configure the DQN agent with suitable hyperparameters, such as learning rate, discount factor, and exploration rate.
• Train the DQN agent using the Gym environment, collecting experience tuples and updating the Q-network weights.
• Evaluate the trained agent's performance by testing it on multiple episodes and measuring the average reward achieved.
• Save and load the trained agent's weights for future use or further training.

[ayush-09]

@adithya-s-k Please assign this to me

[ayush-09]

@miraj0507 Please assign this to me