Plenary Lecture
Monte Carlo Tree Search in Adversarial Environments
Professor Jacek Mańdziuk
Faculty of Mathematics and Information Science
Warsaw University of Technology
Poland
E-mail: J.Mandziuk@mini.pw.edu.pl
Abstract: Monte Carlo Tree Search (MCTS) is a recently proposed search method that gradually builds the problem (search) tree and provides the estimates of nodes’ assessments by means of massive random sampling in the search space. The method gained a lot of interest mainly due to its highly successful application to the game of Go, which is resistant to the classical approaches as it lacks easily computable and compact assessment function. Since its first announcements in 2006 and initial applications to Go, MCTS has proven beneficial also in other domains including Partially Observable Markov Decision Processes, Connection Games, General Game Playing, Constraint Satisfaction Problems or Scheduling. Each MCTS simulation can essentially be divided into two phases: the tree search phase, in which nodes are selected according to a certain policy, and the playout phase in which a random simulation is performed from the leaf of the currently maintained problem tree until the terminal state where the goal value is obtained and back-propagated along the path to the root node. The talk will discuss certain extensions to the canonical MCTS formulation and its default simulation policy in adversarial environments with particular emphasis on increasing the method’s adaptability to variable behavioral pattern of the opponent.
Brief Biography of the Speaker: Prof. Jacek Mańdziuk, Ph.D., D.Sc., received M.Sc. (Honors) and Ph.D. in Applied Mathematics from the Warsaw University of Technology (WUT), Poland in 1989 and 1993, resp. and D.Sc. degree in Computer Science from the Polish Academy of Sciences in 2000. In 2011 he was awarded the title of Full Professor (Professor Titular). He is an Associate Professor at the Faculty of Mathematics and Information Science, WUT, the Head of Division of Artificial Intelligence and Computational Methods, and the Head of Doctoral Programme in Computer Science, at this faculty. He is the author of 3 books (including Knowledge-free and Learning-based Methods in Intelligent Game Playing, Springer, 2010) and 130+ research papers. He served as Program Committee Member for 100+ international conferences. Recently he has been the organizer and Chair of the IEEE SSCI Symposium on Computational Intelligence for Human-like Intelligence (Singapore 2013, Orlando 2014, Cape Town 2015, Athens 2016, Honolulu 2017). He serves/served as an Associate Editor of the IEEE Transactions on Neural Networks and Learning Systems, and the IEEE Transactions on Computational Intelligence and AI in Games. He is a Senior Member of IEEE (M’05, SM’10) and a founding chair of the IEEE ETTC Task Force on Towards Human-like Intelligence. Prof. Mańdziuk was a recipient of the Fulbright Senior Advanced Research Award (UC Berkeley and ICSI Berkeley, USA) and the Robert Schuman Foundation Fellowship (CNRS, Besacon, France). Recently he has been a visiting professor in the School of Computer Engineering, Nanyang Technological University in Singapore (2015-2017). He was also a visiting professor at the University of New South Wales (Australia, 2013), Yonsei University (South Korea, 2011) and the University of Alberta (Canada, 2011). His research interests include application of Computational Intelligence and Artificial Intelligence methods to games, dynamic optimization problems, human-machine cooperation and financial modeling. He is also interested in development of general-purpose human-like learning and problem-solving methods which involve intuition, creativity and multitasking. For more information please visit http://www.mini.pw.edu.pl/~mandziuk