CS 4300: Artificial Intelligence

Assignment: Adversarial Search Agent

Create an adversarial search agent for PettingZoo: Connect Four connect_four.py.

You are welcome to create as many agents as you want, play them against each other, and submit the best one to be graded. You’re also welcome to scrimmage against other students in the class.

Required Process/Files

Use the GitHub repository available for this course to store your solution. Make a directory named connect-four-adversarial. Store the best agent you create in connect-four-adversarial/connect_four_agent.py, with the class ConnectFourAgent that has the following methods:

def __init__(self): initializes the agent
def reset(self): resets the agent for a new game
def agent_function(self, state, player): returns the action that player should take in state. state is the observation from petting zoo. player is the string that identifies which player the agent is.

Your agent is not allowed to use more than 5*60 = 300 seconds total for one game.

You should create a model class for the environment that supports the methods necessary for the mini-max algorithm. It is required that your agent uses mini-max or alpha-beta pruning to choose an action.

Report

The report should describe your board evaluation function, whether you are using mini-max or alpha-beta search, the depth of your cutoff, and list the various attempts you have made.

Required Submissions

Code submitted to github.
A PDF file containing your report submitted to Canvas.

Hints and Resources

connect_four_runner.py

Last Updated 10/08/2024