Scribd
Upload
73 views

Assignment

AlphaGo is a computer program developed by Google in 2015 that plays the board game Go. It is the first computer program to defeat a professional human Go player and world champion. AlphaGo uses artificial neural networks and machine learning techniques like supervised learning from human games and self-play games to evaluate board positions and select moves. It also uses Monte Carlo simulation to search through values from its neural networks to select the next move, allowing it to achieve over a 98% winning rate against other Go players by reducing the search space.

Uploaded by

Abdullah Ammar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
73 views

Assignment

AlphaGo is a computer program developed by Google in 2015 that plays the board game Go. It is the first computer program to defeat a professional human Go player and world champion. AlphaGo uses artificial neural networks and machine learning techniques like supervised learning from human games and self-play games to evaluate board positions and select moves. It also uses Monte Carlo simulation to search through values from its neural networks to select the next move, allowing it to achieve over a 98% winning rate against other Go players by reducing the search space.

Uploaded by

Abdullah Ammar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Artificial Intelligence

CSC 411

Assignment # 02

Abdur Rehman Anwar (01-134142-199)


Suneel Kumar (01-134142-202)
BS(CS)-7A

Department of Computer Science


BAHRIA UNIVERSITY, ISLAMABAD
Artificial Intelligence game AlphaGo
It is board game Go. It was developed by Google corporation in 2015. AlphaGo has 19x19 board size. It
is first computer program to defeat a professional human Go player even the Go world champion.

Artificial Intelligence techniques and Algorithms used in AlphaGo Game

Due to large branching factor, It is useless to used traditional AI techniques such alpha-beta pruning,
tree traversal and heuristic search.
It is used artificial neural network and machine learning techniques to learn from previous and current
knowledge.

I. Artificial Neural Network in AphaGo Game


AlphaGo game uses neural network to evaluate board positions and policy networks to select
moves. This neural network is trained by novel combination of supervised machine learning
from human expert games and self-play games.

II. Monte Carlo Simulation in AlphaGo Game


It is used to search values and policy network to select next move in AlphaGo Game. Due to this
search algorithm, AlphaGo achieved more than 98 % winning rate against any other Go Player.

Conclusion
The highest level goal for AlphaGo is effectiveness in reducing search space, to the number of possible
moves which are computable in reasonable time.

You might also like