Backgammon
Backgammon
Group 1:
- Remco Bras
- Tim Beyer
- Maurice Hermans
- Esther Verhoef
- Thomas Acker
Introduction to backgammon
Demonstration
Game complexity
Different AI algorithms
Future plans
Overview
Overview
Two players
Pieces are moved according to the roll of dice
One player moves clockwise, other
counterclockwise
Pieces kicked off will be placed on the baulk
Scores when reached
Home Board
Introduction
to
backgammon
Introduction
to
backgammon
Let’s demonstrate!
Demonstration
Demonstration
Game
Complexity
Game
Complexity
Neural Network
◦
The state of the board as input
layer
◦
The evaluation function as
hidden layer
◦
The “best move” as output layer
AI
Algorithms
AI
Algorithms
Reinforcement Learning (RL)
◦
Learning by playing matches against itself
◦
Temporal Difference Learning (TDL)
Changes after every time step
Temporal
Changes through differences
Difference
Learning of a evaluation (value) function
Learning
AI
Algorithms
AI
Algorithms
Pubeval
◦
A benchmark “player” by Gerry Tesauro
◦
Fixed weights
Fixed strategy
◦
Used very often to compare different approaches
◦
Used for illustrating training effects
AI
Algorithms
AI
Algorithms
Implement Neural networks
Vary training sessions
Future
Plans
Future
Plans
Week1
Week2
Week3
Week4
Week5
Week6
Week7
Research
Implement
AI
Testing and
debugging
Improve
AI
Working
on
presentation
Gannt-Chart