[computer-go] FW: Microsoft Research Lectures; Erik Van der Werf; Maastricht University

[Jack]

FYI


-----Original Message-----
From: Sarah Cater (MSRC) [mailto:sarahc at microsoft.com] 
Sent: 20 November 2006 11:13
To: Jack Lang
Subject: Microsoft Research Lectures; Erik Van der Werf; Maastricht
University

MICROSOFT RESEARCH LECTURE
This is a PUBLIC lecture
________________________________________
TITLE: AI techniques for the game of Go
SPEAKER: Erik Van der Werf
INSTITUTION: Maastricht University/GN Resound
HOST: Thore Graepel
DATE: 27 November 2006
TIME: 15:00 - 16:00
MEETING ROOM: Lecture-room small(50)
ADDRESS: Microsoft Research Ltd, 7 J J Thomson Avenue (Off Madingley Road),
Cambridge
Since the founding years of Artificial Intelligence (AI) computer games have
been used as a testing ground for AI algorithms. The game of Go is by far
the most complex popular board game in the class of two-player
perfect-information games. Despite significant research efforts over the
last four decades, Go programs still play at only a moderately weak amateur
level. Now, after the recent succumb in Chess, Go has become the task par
excellence in game AI research. In this talk I will present an overview of
my Ph.D. research in computer Go performed between 2000 and 2005 in the
computer games group at Maastricht University. The research focused on: (1)
searching techniques, which were already successful in games such as Chess,
and (2) learning techniques, from pattern recognition and machine learning,
which were already successful in various other complex domains. We have
applied learning techniques successfully to various Go-related tasks such as
move prediction, score prediction, classifying life and death, and
estimating potential territory. The searching techniques were used to build
a strong program for the capture game (a simplified version of Go), solve it
up to size 6x6, and create the first program in the world to have solved Go
on the 5x5 board (which still holds the record as the largest board solved
by computer). Finally, all techniques have been integrated in the Go program
Magog, which enabled it to win the bronze medal in the 9th Computer
Olympiad.


---
You are currently subscribed to msrclectures as: jack at lang.net
To unsubscribe send a blank email to
leave-msrclectures-8707F at list.research.microsoft.com
Please see our Privacy Statement:
http://list.research.microsoft.com/privacy/privacy.htm