The game of Go has long been viewed as the most challenging of classic games for artificial intelligence owing to its enormous search space and the difficulty of evaluating board positions and moves. Here we introduce a new approach to computer Go that uses ‘value networks’ to evaluate board positions and ‘policy networks’ to select moves. These deep neural networks are trained by a novel combination of supervised learning from human expert games, and reinforcement learning from games of self-play. Without any lookahead search, the neural networks play Go at the level of stateof-the-art Monte Carlo tree search programs that simulate thousands of random games of self-play. We also introduce a new search algorithm that combines Monte Carlo simulation with value and policy networks. Using this search algorithm, our program AlphaGo achieved a 99.8% winning rate against other Go programs, and defeated the human European Go champion by 5 games to 0. This is the first time that a computer program has defeated a human professional player in the full-sized game of Go, a feat previously thought to be at least a decade away.

/pdf/mastering-the-game-of-go-with-deep-neural-networks-and-tree-3viuxo6uxm.pdf

Mastering the game of Go with deep neural networks and tree search

Recent progress in artificial intelligence has renewed interest in building systems that learn and think like people. Many advances have come from using deep neural networks trained end-to-end in tasks such as object recognition, video games, and board games, achieving performance that equals or even beats that of humans in some respects. Despite their biological inspiration and performance achievements, these systems differ from human intelligence in crucial ways. We review progress in cognitive science suggesting that truly human-like learning and thinking machines will have to reach beyond current engineering trends in both what they learn and how they learn it. Specifically, we argue that these machines should (1) build causal models of the world that support explanation and understanding, rather than merely solving pattern recognition problems; (2) ground learning in intuitive theories of physics and psychology to support and enrich the knowledge that is learned; and (3) harness compositionality and learning-to-learn to rapidly acquire and generalize knowledge to new tasks and situations. We suggest concrete challenges and promising routes toward these goals that can combine the strengths of recent neural network advances with more structured cognitive models.

/pdf/building-machines-that-learn-and-think-like-people-w8p3qwruiy.pdf

Building machines that learn and think like people.

Monte-Carlo Tree Search (MCTS) is a new best-first search method that started a revolution in the field of Computer Go. Parallelizing MCTS is an important way to increase the strength of any Go program. In this article, we discuss three parallelization methods for MCTS: leaf parallelization, root parallelization,and tree parallelization. To be effective tree parallelization requires two techniques: adequately handling of (1) local mutexes and (2) virtual loss. Experiments in 1313 Go reveal that in the program Mango root parallelization may lead to the best results for a specific time setting and specific program parame- ters. However, as soon as the selection mechanism is able to handle more adequately the balance of exploitation and exploration, tree paralleliza- tion should have attention too and could become a second choice for parallelizing MCTS. Preliminary experiments on the smaller 99 board provide promising prospects for tree parallelization.

Parallel Monte-Carlo tree search

Mastering the game of Go has remained a longstanding challenge to the field of AI. Modern computer Go programs rely on processing millions of possible future positions to play well, but intuitively a stronger and more 'humanlike' way to play the game would be to rely on pattern recognition rather than brute force computation. Following this sentiment, we train deep convolutional neural networks to play Go by training them to predict the moves made by expert Go players. To solve this problem we introduce a number of novel techniques, including a method of tying weights in the network to 'hard code' symmetries that are expected to exist in the target function, and demonstrate in an ablation study they considerably improve performance. Our final networks are able to achieve move prediction accuracies of 41.1% and 44.4% on two different Go datasets, surpassing previous state of the art on this task by significant margins. Additionally, while previous move prediction systems have not yielded strong Go playing programs, we show that the networks trained in this work acquired high levels of skill. Our convolutional neural networks can consistently defeat the well known Go program GNU Go and win some games against state of the art Go playing program Fuego while using a fraction of the play time.

/pdf/training-deep-convolutional-neural-networks-to-play-go-epagdbutdn.pdf

Training Deep Convolutional Neural Networks to Play Go

The game of Go is more challenging than other board games, due to the difficulty of constructing a position or move evaluation function. In this paper we investigate whether deep convolutional networks can be used to directly represent and learn this knowledge. We train a large 12-layer convolutional neural network by supervised learning from a database of human professional games. The network correctly predicts the expert move in 55% of positions, equalling the accuracy of a 6 dan human player. When the trained convolutional network was used directly to play games of Go, without any search, it beat the traditional search program GnuGo in 97% of games, and matched the performance of a state-of-the-art Monte-Carlo tree search that simulates a million positions per move.

Move Evaluation in Go Using Deep Convolutional Neural Networks

We present a state of the art implementation of the Monte Carlo Tree Search algorithm for the game of Go. Our Pachi software is currently one of the strongest open source Go programs, competing at the top level with other programs and playing evenly against advanced human players. We describe our implementation and choice of published algorithms as well as three notable original improvements: (1) an adaptive time control algorithm, (2) dynamic komi, and (3) the usage of the criticality statistic. We also present new methods to achieve efficient scaling both in terms of multiple threads and multiple machines in a cluster.

Jean-loup Gailly

Papers

PACHI: State of the Art Open Source Go Program