Tobias+Graf

a German computer scientist affiliated with the University of Paderborn. He holds a B.Sc. degree in 2010 on the topic of parallel UCT with high-performance computing, and a M.Sc. in 2012, elaborating on adaptiv playouts in Monte-Carlo Tree Search applied to Computer Go.
 * Home * People * Tobias Graf**
 * [[image:TobiasGraf.jpg link="http://www.uni-paderborn.de/mitteilung/143732/"]] ||~ || **Tobias Graf**,

He is auhor of the Go playing program [|Abakus], winning 9x9 and 13x13, and runner-up in 9x19 at the 19th Computer Olympiad, and silver and bronze at the 18th Computer Olympiad, and before collaborated with Lars Schaefers on the Go playing program [|Gomorra]. During the Computer and Games conference, 2013 in [|Yokohama], he lectured on detection of [|capturing races (Semeai)] in Go, co-authored by Lars Schaefers and Marco Platzner. toc =Selected Publications=
 * Tobias Graf ||~ ||^ ||
 * Tobias Graf ||~ ||^ ||
 * Tobias Graf (**2010**). //Parallelization of the UCT Algorithm on HPC-Clusters//. Bachelor's Thesis, University of Paderborn
 * Tobias Graf, Ulf Lorenz, Marco Platzner, Lars Schaefers (**2011**). //Parallel Monte-Carlo Tree Search for HPC Systems//. [|Euro-Par 2011], [|pdf]
 * Tobias Graf (**2012**). //Adaptive Playouts in der Monte-Carlo Spielbaumsuche am Anwendungsfall Go//. Master's Thesis, University of Paderborn (German)
 * Tobias Graf, Lars Schaefers, Marco Platzner (**2013**). //On Semeai Detection in Monte-Carlo Go//. CG 2013, [|pdf]
 * Tobias Graf, Marco Platzner (**2015**). //Adaptive Playouts in Monte Carlo Tree Search with Policy Gradient Reinforcement Learning//. Advances in Computer Games 14
 * Tobias Graf, Marco Platzner (**2016**). //Using Deep Convolutional Neural Networks in Monte Carlo Tree Search//. CG 2016

=External Links=
 * [|Tobias Graf's ICGA Tournaments]
 * [|Universität Paderborn: Silbermedaille und „Best Paper Award“ bei Computer-Olympiade in Yokohama], October 24, 2013 (German)

=References= =What links here?= include page="Tobias Graf" component="backlinks" limit="40"
 * Up one Level**