Table of Contents Link to heading

• Summary
• Files Overview
• Heuristic Score Calculation
• Minimax Algorithm
  • Key Parts
  • Methods
• GUI Features
• Dependencies
• Demo Video

Summary Link to heading

This project implements an AI agent for Connect Four using the Minimax algorithm with alpha-beta pruning. The game features a graphical user interface (GUI) where users can play against the AI, adjust the search depth, and toggle pruning.

Files Overview Link to heading

• GUI.java: Handles the game interface, user input, and display. It interacts with the Minimax AI to determine the AI’s moves.
• Minimax.java: Contains the Minimax algorithm implementation, including heuristic evaluation and alpha-beta pruning.

Heuristic Score Calculation Link to heading

The heuristic function evaluates the board state by analyzing all possible sequences of four consecutive cells (horizontal, vertical, diagonal). For each sequence:

1. Count AI agent a and opponent o pieces in the sequence.
2. Score adjustment:
   • If the sequence contains only AI pieces:
     score += 10^(number of AI pieces)
   • If the sequence contains only opponent pieces:
     score -= 10^(number of opponent pieces)
   • Mixed/empty sequences do not affect the score.

This prioritizes creating/blocking potential wins and rewards the AI for controlling more cells in a sequence.

Minimax Algorithm Link to heading

Key Parts Link to heading

• State Class: Represents a game state with:

  • boardState: 42-character string (a, o, # for empty).
  • depth: Current depth in the search tree.
  • maxOrMin: Indicates if it’s the AI’s (maximizer) or opponent’s (minimizer) turn.

• Algorithm Flow:

  1. Maximizer: Selects the move with the highest heuristic value from successors.
  2. Minimizer: Selects the move with the lowest heuristic value from successors.
  3. Depth Limiting: Stops recursion at a user-defined depth (default: 2) and returns the heuristic score.

• Alpha-Beta Pruning: Optimizes Minimax by pruning branches that cannot influence the final decision, reducing computation time.

Methods Link to heading

• value(State state): Entry point for Minimax, delegates to maximizer() or minimizer().
• abValue(State state, int alpha, int beta): Alpha-beta variant of value().
• getSuccessors(State state): Generates valid next states for all columns.

GUI Features Link to heading

• Interactive Board: Click “Drop” buttons to place pieces.
• Settings:
  • Depth Adjustment: Controls how many moves ahead should the AI plans (2–7).
  • Pruning Toggle: Enables/disables alpha-beta pruning for faster decisions.
• Win Detection: Displays scores and results when the board is full.

Dependencies Link to heading

• Java Swing for GUI components.
• Image icons for board visuals (provided in the src/ directory).

Demo Video Link to heading