https://github.com/malintha-induwara/connect-four-game

A Connect Four game implemented in Java using the Monte Carlo algorithm for AI decision making.
https://github.com/malintha-induwara/connect-four-game

java java-fx monte-carlo-simulation oop

Last synced: 4 months ago
JSON representation

A Connect Four game implemented in Java using the Monte Carlo algorithm for AI decision making.

• Host: GitHub
• URL: https://github.com/malintha-induwara/connect-four-game
• Owner: malintha-induwara
• License: mit
• Created: 2023-09-27T15:11:25.000Z (about 2 years ago)
• Default Branch: main
• Last Pushed: 2025-04-16T19:48:36.000Z (6 months ago)
• Last Synced: 2025-06-13T08:02:47.573Z (4 months ago)
• Topics: java, java-fx, monte-carlo-simulation, oop
• Language: Java
• Homepage:
• Size: 130 KB
• Stars: 3
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.txt

Awesome Lists containing this project

README

          
# Connect Four with Monte Carlo Algorithm.

* Connect 4 game made using JavaFX and Monte Carlo Tree Search Algorithem As AI.

### Create Game

![5](https://github.com/malintha-induwara/connect-four-game/assets/60071404/0c648510-9e3a-47c9-b7c9-bc5a405721f3)

### Game Progression

![2](https://github.com/malintha-induwara/connect-four-game/assets/60071404/1b969912-d73a-4ecd-8620-5642e4fa9a1a)

![3](https://github.com/malintha-induwara/connect-four-game/assets/60071404/87b008d2-2db9-4402-a7b8-3d74c3bf2163)

![4](https://github.com/malintha-induwara/connect-four-game/assets/60071404/c527d249-fdee-465a-82a3-4adaae67df89)

### AI Implementation & LLM Testing

This project has also been used as a benchmark for testing the ability of large language models (LLMs) to implement Monte Carlo Tree Search (MCTS). Since GPT-3, all (free-tier) models from OpenAI have struggled to produce a working implementation of MCTS, including other models such as Claude 3.5 (Anthropic) and DeepSeek-R1 (DeepSeekAI). However, o3-Mini(Low) successfully implemented the algorithm, making it the first in this series of models to do so.

You can find the LLM produced code inside the folder **`src/main/java/lk/ijse/dep/service/llm`**.

The implementation inside the `AiPlayer` class is the original implementation and was developed entirely without any help from LLMs.

### How to use this repo

* Open the pom.xml via IntelliJ IDEA

* Make sure to the open it as a project, if prompt

* Reload the pom.xml file via Maven Tool Window

* Create a run configuration for Maven via Run > Edit Configuration

* Add javafx:run as the Run command

  

That's it.