Backgammon AI

// alpha-beta pruning & expectimax

What is Backgammon & Why AI?

One of the oldest board games in existence (5000 years old!), backgammon combines elements of strategy, probability, and skill. Players, and in this case AI agents, compete to move their pieces around the board, trying to remove all pieces before the opponent. Crafting an AI to play Backgammon is a captivating challenge due to the game's intricate blend of tactics and chance. This project was developed in Python with a partner, and implements Expectimax and Alpha-Beta Pruning algorithms to create two intelligent agents to face off against one another. This project allowed me to sharpen and expand my knowledge & skills in:

Python Programming

Game Theory & Strategy

AI Search Algorithms

Collaboration & Communication

Learning Search Algorithms

This project was part of my CSE 415, Intro to AI course. While this wasn't the most complicated or difficult project I completed in this class, it is a story of perserverence and adaptation. Starting with BFS and DFS, we learned the most common and simplest search algorithms. But as we introduced heuristics and other optimizations, we could more efficiently "solve" problems. Using these newly learned concepts, my teammate and I started our work on two backgammon agents. One utilizing a minimax algorithms with alpha-beta pruning and expectimax on the other. However, we soon ran into some problems...

Timeline

① Exploring Course Content

We delved into adversarial search with and without stochasticity. We learned key strategies like minimax for deterministic games, alpha-beta pruning for efficiency, and expectimax for handling games with randomness.

② Stumbling Blocks

After reviewing class notes and slides, I began implementing my algorithms. However, quickly I ran into a wall. I had created too many bugs and errors, ultimately leading me to be very confused and unable to progress further.

③ (Mis)communication

After encountering implementation issues, I sought collaboration with my partner, but surprisingly received no response. Determined to overcome the hurdles, I turned to office hours for guidance, driven by my commitment to find solutions and achieve our project goals.

④ Perserverence

Ultimately, the due date came and went. I learned my partner had gotten quite sick and couldn't work on the project. Still we we determined to finish our project, regardless of any point deductions. The next few days we got back on track, worked hard and completed our backgammon agents.

Finishing Touches

While I gained insights from office hours and class materials, I failed to piece everything together before the deadline. Frustration was really kicking in and I even thought about dropping the class at this point, but I remained resolute. Determined to complete the project, we dedicted the next three days to communication, brainstorming, debugging. In the end, seeing our AI agents play a game of Backgammon, employing the algorithms we had struggled to implement, was immensely satisfying and a testament to our perseverance and teamwork.

Here's the Github Repo

Message Me!

First Name *

Last Name

Email *

Message * * required