The Go-Playing Program Called Go81
نویسنده
چکیده
Go is an ancient game, for which it has proven to be very difficult to create an artificial player. Go81 is yet another try in that direction. The main idea is as follows: firstly, create a so called ant that tries to play as well as possible, given that it has to be very fast and slightly randomized. Secondly, use these ants to play the game from the current state to the end several times and make use of the information from these possible futures. This approach avoids the evaluation of an unfinished game, which is perhaps the one thing that makes computer Go so difficult. Two versions of Go81, one for Palm and one for a Linux console, are tested against a shareware program AIGO for Palm and an open source project GNU Go accordingly. The Palm version is as strong as AIGO and the console version is two stones weaker than GNU Go on a 9 by 9 board. The proposed approach can also be used to generate interesting data to be studied with machine learning techniques.
منابع مشابه
Multi-Labelled Value Networks for Computer Go
This paper proposes a new approach to a novel value network architecture for the game Go, called a multi-labelled (ML) value network. In the ML value network, different values (win rates) are trained simultaneously for different settings of komi, a compensation given to balance the initiative of playing first. The ML value network has three advantages, (a) it outputs values for different komi, ...
متن کاملThe Integration of A Priori Knowledge into a Go Playing Neural Network
The best current computer Go programs are hand crafted expert systems. They are using conventional AI technics such as pattern matching, rule based systems and goal oriented selective search. Due to the increasing complexity of managing this kind of knowledge representation by hand, the playing strength of these programs is still far from human master level. This article describes methods for i...
متن کاملUsing Hard and Soft Artificial Intelligence Algorithms to Simulate Human Go Playing Techniques
We describe the development of a Go playing computer program that combines the use of hard Artificial Intelligence (AI) techniques (alpha-beta search trees) with soft AI techniques (neural networks). The concept is based on a model of human play where selection of plausible moves is made using a gestalt process based on experience and the plausible moves are subjected to an objective analysis. ...
متن کاملConvolutional Monte Carlo Rollouts in Go
In this work, we present a MCTS-based Go-playing program which uses convolutional networks in all parts. Our method performs MCTS in batches, explores the Monte Carlo search tree using Thompson sampling and a convolutional network, and evaluates convnet-based rollouts on the GPU. We achieve strong win rates against open source Go programs and attain competitive results against state of the art ...
متن کاملHonte, a Go-Playing Program Using Neural Nets
The go-playing program Honte is described. It uses neural nets together with more conventional AI-methods like alpha-beta search. A neural net is trained by supervised learning to imitate local shapes made in a database of expert games. A second net is trained to estimate the safety of groups by self play using TD(λ)learning. A third net is trained to estimate territorial potential of unoccupie...
متن کامل