{"type":"dissertation","date_updated":"2022-01-06T06:50:50Z","page":"133","publication_status":"published","supervisor":[{"id":"398","last_name":"Platzner","full_name":"Platzner, Marco","first_name":"Marco"}],"language":[{"iso":"eng"}],"date_created":"2019-07-10T11:58:06Z","user_id":"3118","citation":{"short":"L. Schäfers, Parallel Monte-Carlo Tree Search for HPC Systems and Its Application to Computer Go, Logos Verlag Berlin GmbH, Berlin, 2014.","chicago":"Schäfers, Lars. <i>Parallel Monte-Carlo Tree Search for HPC Systems and Its Application to Computer Go</i>. Berlin: Logos Verlag Berlin GmbH, 2014.","bibtex":"@book{Schäfers_2014, place={Berlin}, title={Parallel Monte-Carlo Tree Search for HPC Systems and its Application to Computer Go}, publisher={Logos Verlag Berlin GmbH}, author={Schäfers, Lars}, year={2014} }","apa":"Schäfers, L. (2014). <i>Parallel Monte-Carlo Tree Search for HPC Systems and its Application to Computer Go</i>. Berlin: Logos Verlag Berlin GmbH.","mla":"Schäfers, Lars. <i>Parallel Monte-Carlo Tree Search for HPC Systems and Its Application to Computer Go</i>. Logos Verlag Berlin GmbH, 2014.","ama":"Schäfers L. <i>Parallel Monte-Carlo Tree Search for HPC Systems and Its Application to Computer Go</i>. Berlin: Logos Verlag Berlin GmbH; 2014.","ieee":"L. Schäfers, <i>Parallel Monte-Carlo Tree Search for HPC Systems and its Application to Computer Go</i>. Berlin: Logos Verlag Berlin GmbH, 2014."},"title":"Parallel Monte-Carlo Tree Search for HPC Systems and its Application to Computer Go","_id":"10733","department":[{"_id":"78"}],"publication_identifier":{"isbn":["978-3-8325-3748-7"]},"author":[{"first_name":"Lars","full_name":"Schäfers, Lars","last_name":"Schäfers"}],"abstract":[{"lang":"eng","text":"Monte-Carlo Tree Search (MCTS) is a class of simulation-based search algorithms. It brought about great success in the past few years regarding the evaluation of deterministic two-player games such as the Asian board game Go.\r\n\r\nIn this thesis, we present a parallelization of the most popular MCTS variant for large HPC compute clusters that efficiently shares a single game tree representation in a distributed memory environment and scales up to 128 compute nodes and 2048 cores. It is hereby one of the most powerful MCTS parallelizations to date.\r\n\r\nIn order to measure the impact of our parallelization on the search quality and remain comparable to the most advanced MCTS implementations to date, we implemented it in a state-of-the-art Go engine Gomorra, making it competitive with the strongest Go programs in the world.\r\n\r\nWe further present an empirical comparison of different Bayesian ranking systems when being used for predicting expert moves for the game of Go and introduce a novel technique for automated detection and analysis of evaluation uncertainties that show up during MCTS searches."}],"publisher":"Logos Verlag Berlin GmbH","year":"2014","place":"Berlin","status":"public"}