Othello Artificial Intelligence With Machine Learning

1
Othello Artificial IntelligenceWith Machine
Learning

• Computer Systems TJHSST
• Nick Sidawy

2
Introduction

• Machine learning is an extensive filed of study.
  Most of what is done with machine learning is
  tied to artificial intelligence which is why
  Othello seemed to be a good vessel for my
  research. It is a simple enough game for me to
  work on, yet difficult enough to keep me working
  throughout the quarters.

3
Purpose

• The purpose of this research project is to
  implement machine learning with artificial
  intelligence. The reason for this is two-fold
• First, to create a very effective Othello AI.
• Second, and more academically oriented, is to
  gain a deeper understanding of machine learning.

4
Goal Breakdown

• The first goal i would like to achieve with this
  project is to program an effective
  forward-checker for the AI.
• The second goal is to use a genetic algorithm to
  formulate the best evaluation function for the AI
  to use.
• My last goal is to have the AI learn from each
  move it makes so that the more it plays, the
  faster and better it will perform.

5
Development

• The project will be coded in java.
• I have separated it into 5 different programs.
  The driver, the panel (which controls the game),
  the AI, the game functions (such as capturing
  pieces) and a class that I have named MyButton.

6
Construction

• There are two algorithms that this quarter was
  focused on
• The forward-checking algorithm (Minimax)
• The evaluation function
• A MyButton class was also created

7
MyButton Class

• The MyButton class extends Jbutton and has the
  following variations
• It has a paint component
• It can store values which indicate if the spot on
  the board is occupied by a piece and will draw
  images accordingly

8
Forward-Checker

• The goal of a forward Checker is to traverse a
  tree of possible moves and picking the move that
  will lead to the best scenario down the line.
  The ply determines how many levels of the tree it
  goes through.
• The trick is that at each level of the three it
  picks the move that is best for the player it is
  simulating for. Therefore, the computer assumes
  the opponent will play perfectly.
• It is nicknamed the Minimax algorithm for this
  reason. (See Diagram A)

9
(No Transcript)
10
Evaluation Function

• This function returns a number rating how good a
  particular board is for a player. It does this
  based on the positions of the pieces and amount
  of available moves for each player. For example,
  pieces in the corners are very valuable and will
  add many more points to the rating than a piece
  near the center.

11
Conclusions

• Thus far into the project I have learned the
  importance of the evaluation function. A strong
  evaluation function has much more weight than a
  forward-checker with a high-ply.
• When the project is complete I plan on having a
  very skilled Othello AI that uses machine
  learning extensively and will be challenging for
  the best players.