Evolving Monte-Carlo Tree Search Algorithms
نویسنده
چکیده
Monte-Carlo tree search is a recent and powerful algorithm that has been applied with success to the game of Go. Combining it with heuristics for the early development of nodes with few samples makes it even stronger. We use Genetic Programming to evolve such heuristics. The heuristics discovered by Genetic Programming outperform UCT and RAVE. Our Genetic Programming system computes the fitness using a Swiss tournament, and it selects valid individuals.
منابع مشابه
Migrating Birds Optimization: A New Meta-heuristic Approach and Its Application to the Quadratic Assignment Problem
A Study on the Mutation Rates of a Genetic Algorithm Interacting with a Sandpile p. 32 On the Design of Boolean Network Robots p. 43 EvoGames Contributions Multiple Tree for Partially Observable Monte-Carlo Tree Search p. 53 Evolving Interesting Maps for a First Person Shooter p. 63 Revisiting Monte-Carlo Tree Search on a Normal Form Game: NoGo p. 73 Nested Look-Ahead Evolutionary Algorithm Bas...
متن کاملFuegito: an Educational Software Package for Game Tree Search
Fuegito is an educational software package for learning about programming two player games. The package provides a simple, yet flexible and extensible framework which allows students to study the core search algorithms of computer game-playing, and extend them easily in projects. The current version of Fuegito supports alphabeta and Monte Carlo Tree Search as main game tree search methods, as w...
متن کاملState Aggregation in Monte Carlo Tree Search
Monte Carlo tree search (MCTS) algorithms are a popular approach to online decision-making in Markov decision processes (MDPs). These algorithms can, however, perform poorly in MDPs with high stochastic branching factors. In this paper, we study state aggregation as a way of reducing stochastic branching in tree search. Prior work has studied formal properties of MDP state aggregation in the co...
متن کاملA Lock-Free Multithreaded Monte-Carlo Tree Search Algorithm
With the recent success of Monte-Carlo tree search algorithms in Go and other games, and the increasing number of cores in standard CPUs, the efficient parallelization of the search has become an important issue. We present a new lock-free parallel algorithm for Monte-Carlo tree search which takes advantage of the memory model of the IA-32 and Intel-64 CPU architectures and intentionally ignore...
متن کاملA General Multi-Agent Modal Logic K Framework for Game Tree Search
We present an application of Multi-Agent Modal Logic K (MMLK) to model dynamic strategy game properties. We also provide several search algorithms to decide the model checking problem in MMLK. In this framework, we distinguish between the solution concept of interest which is represented by a class of formulas in MMLK and the search algorithm proper. The solution concept defines the shape of th...
متن کامل