Tristan Cazenave
Tristan Cazenave
Member of the editorial board of the Revue d'Intelligence Artificielle
Special volume of the Revue d'Intelligence Artificielle
Program committee member of CG 2008, ICHSL.6, AAAI-07, CGW 2007, H2PTM 2007, ECAI 2006, CG'2006, H2PTM'05, ACG 11, AAAI-04, CG'04, CG'98
Reviewer for Artificial Intelligence, JAIR, Journal of Computer Science and Technology, ICGA Journal, Theoretical Computer Science, Information Sciences, Computational Intelligence, New Mathematics and Natural Computation, Integers, Revue d'Intelligence Artificielle, WCCI 2008, IJCAI-07, IEEE CIG-07, IPMU-06, ACG 10, ECAI-02, CG 2002, CG 2000
In 1991, I graduated from INT telecommunication engineer school.
In 1992, I co-founded with Jean-Sebastien Hongre and Frederic Benichou a company which later became Planete Interactive, a web agency employing 250 people in 2007.
In 1996, I defended my PhD thesis at Paris 6 University on a system learning to play games by observing itself.
In 1997, I wrote a meta-program that wrote a Go program that finished 5th out of 40 Go programs in the international FOST cup competition, held at Nagoya during IJCAI-97.
In April 2000, as I wrote the Monte-Carlo part of our computer Go survey for the Artificial Intelligence journal, and following a discussion on Monte-Carlo Go with Bruno Bouzy, I re-implemented Bernd Bruegmann's Gobble and compared it to a more simple Monte-Carlo sampling approach, finding only little difference in level. The idea was further developed by Bernard Helmstetter, Bruno Bouzy, Guillaume Chaslot, Remi Coulom, Yizao Wang and Sylvain Gelly who all much improved Monte-Carlo Go to the point it has become the current best approach to computer Go.
In 2002, one of my threat-based search algorithms solved John Conway's Phutball for size 9x9 and 11x11, as well as 6x6 Atarigo.
In 2003, I defended my habilitation thesis at Paris 8 University.
In 2005, I wrote a Monte-Carlo based Phantom Go program competitive with human Go players. An improved version won a gold medal at the 2007 computer Olympiad .
In 2006, I wrote a multiple sequence alignment algorithm that finds multiple alignments more than 13,000 times faster and with less than 150 times less memory than the optimal algorithm with low error rates. I also supervised a book on Artificial Intelligence and games.
In 2007, Bernard Helmstetter was my first student to defend his PhD thesis. With the help of Nicolas Jouandeau, I made a parallel version of UCT on a cluster. My reflexive Monte-Carlo program broke world records with solutions of length 76 then 78 to Morpion Solitaire disjoint version. I later matched the new 79 record.
