1 % This file was created with JabRef 2.6.
5 author = {{B
}audi\v
{s
}, {P
}etr and others
},
6 title = {Pachi
--- Simple Go
/Baduk
/Weiqi Bot
},
7 url
= {http
://repo.or.cz
/w
/pachi.git
},
12 author = {Ales Cieply and others
},
13 title = {EGF ratings system
-- System description
},
14 url
= {http
://www.europeangodatabase.eu
/EGD
/EGF_rating_system.php
},
16 timestamp
= {2011.03.10}
20 author = {Ales Cieply and others
},
21 title = {Go Ratings of Professional and Strong Amateur Players
},
22 url
= {http
://www.goweb.cz
/progor
},
26 @ARTICLE
{CoverHart1967
,
27 author = {Thomas M. Cover and Peter E. Hart
},
28 title = {Nearest neighbor pattern classification
},
29 journal = {IEEE Transactions on Information Theory
},
35 timestamp
= {2010.03.10}
38 @INPROCEEDINGS
{PatElo
,
39 author = {{C
}oulom
, {R
}{\'e
}mi
},
40 title = { {C
}omputing
{E
}lo
{R
}atings of
{M
}ove
{P
}atterns in the
{G
}ame of
42 booktitle = {{C
}omputer
{G
}ames
{W
}orkshop
},
44 editor = {van den
{H
}erik
, {H
}.
{J
}aap and
{M
}ark
{W
}inands and
{J
}os
{U
}iterwijk
45 and
{M
}aarten
{S
}chadd
},
46 address = {{A
}msterdam
{P
}ays
-{B
}as
},
47 abstract = {{M
}ove patterns are an essential method to incorporate domain knowledge
48 into
{G
}o
-playing programs.
{T
}his paper presents a new
{B
}ayesian
49 technique for supervised learning of such patterns from game records
,
50 based on a generalization of
{E
}lo ratings.
{E
}ach sample move in
51 the training data is considered as a victory of a team of pattern
52 features.
{E
}lo ratings of individual pattern features are computed
53 from these victories
, and can be used in previously unseen positions
54 to compute a probability distribution over legal moves.
{I
}n this
55 approach
, several pattern features may be combined
, without an exponential
56 cost in the
number of features.
{D
}espite a very small
number of
57 training games
(652), this algorithm outperforms most previous pattern
-learning
58 algorithms
, both in terms of mean log
-evidence
(−
2.69), and prediction
59 rate
(34.9%). {A} 19x19 {M}onte-{C}arlo program improved with these
60 patterns reached the level of the strongest classical programs.
},
61 affiliation
= {{SEQUEL
} - {INRIA
} {F
}uturs
- {INRIA
} - {CNRS
} : {UMR
}8022 - {CNRS
}
62 : {UMR
}8146 - {U
}niversit
{\'e
} des
{S
}ciences et
{T
}echnologies de
63 {L
}ille
- {L
}ille
{I
} - {U
}niversit
{\'e
} {C
}harles de
{G
}aulle
-
64 {L
}ille
{III
} - {E
}cole
{C
}entrale de
{L
}ille
},
65 audience
= {non sp
{\'e
}cifi
{\'e
}e
},
66 url
= {http
://hal.inria.fr
/inria
-00149859/en
/}
70 title = {Pattern classification and scene analysis
},
71 publisher = {A Wiley
-Interscience Publication
, New York
: Wiley
},
73 author = {{Duda
}, R.~O. and
{Hart
}, P.~E.
}
77 title = {The rating of chessplayers
, past and present
},
78 publisher = {Arco
, New York
},
80 author = {Arpad E. Elo
},
86 author = {Gunnar Farneb\"
{a
}ck
},
87 title = {GTP
--- Go Text Protocol
},
88 url
= {http
://www.lysator.liu.se
/~gunnar
/gtp
/},
93 author = {Ulrich G\"ortz
},
94 title = {Kombilo
--- a go database program
},
95 url
= {http
://www.u
-go.net
/kombilo
/},
99 @INPROCEEDINGS
{GellySilver2008
,
100 author = {Gelly
, Sylvain and Silver
, David
},
101 title = {Achieving master level play in
9x9 computer go
},
102 booktitle = {AAAI'
08: Proceedings of the
23rd national conference on Artificial
105 pages = {1537--1540},
106 publisher = {AAAI Press
},
107 abstract = {The UCT algorithm uses Monte
-Carlo simulation to estimate the value
108 of states in a search tree from the current state. However
, the first
109 time a state is encountered
, UCT has no knowledge
, and is unable
110 to generalise from previous experience. We describe two extensions
111 that
address these weaknesses. Our first algorithm
, heuristic UCT
,
112 incorporates prior knowledge in the form of a value function. The
113 value function can be learned offline
, using a linear combination
114 of a million binary features
, with weights trained by temporal
-difference
115 learning. Our second algorithm
, UCT
-RAVE
, forms a rapid online generalisation
116 based on the value of moves. We applied our algorithms to the domain
117 of
9 ×
9 Computer Go
, using the program MoGo. Using both heuristic
118 UCT and RAVE
, MoGo became the first program to achieve human master
119 level in competitive play.
},
120 isbn
= {978-1-57735-368-3},
121 location
= {Chicago
, Illinois
}
124 @ELECTRONIC
{GoDiscThread
,
125 author = {GoDiscussions
},
126 title = {Pro Styles
(discussion thread
)},
127 url
= {http
://www.godiscussions.com
/forum
/showthread.php?t
=10980},
132 author = {Frank de Groot
},
133 title = {Moyo Go Studio
},
134 url
= {http
://www.moyogo.com
/},
138 @ELECTRONIC
{SociomapSite
,
139 author = {{\relax QED
} GROUP
},
140 title = {Sociomapping Suite
},
141 url
= {http
://www.sociomap.com
/},
146 author = {{\relax QED
} GROUP
},
147 title = {Team Profile Analyzer
},
148 url
= {http
://www.teamprofileanalyzer.com
/},
153 author = {Hall
, T. Mark and Fairbairn
, John
},
154 title = {Games of Go on Disk
--- {GoGoD
} Encyclopaedia and Database
},
155 url
= {http
://www.gogod.co.uk
/},
160 title = {Neural Networks
: A Comprehensive Foundation
},
161 publisher = {Macmillan
},
163 author = {Haykin
, S.
},
164 address = {New York
},
165 citeulike
-article
-id
= {2429223},
166 keywords = {neural\_nets
},
168 posted
-at
= {2008-02-26 12:33:02},
170 timestamp
= {2010.06.13}
174 author = {Arno Hollosi
},
175 title = {SGF File Format
},
176 url
= {http
://www.red
-bean.com
/sgf
/},
181 title = {Visualization of Sociomaps
},
182 publisher = {MFF UK
},
184 author = {{H
}\"
{o
}schl
, {C
}yril
},
185 note = {{Bachelor Thesis
}},
189 @MISC
{SociomapsPersonal
,
190 author = {{H
}\"
{o
}schl
, {C
}yril
},
191 howpublished = {personal communication
},
196 author = {Jean
-loup Gailly
, Bill Hosken and others
},
197 title = {The Go Teaching Ladder
},
198 url
= {http
://gtl.xmp.net
/},
203 title = {Principal Component Analysis
},
204 publisher = {Springer
, New York
},
206 author = {I.T. Jolliffe
},
210 @ELECTRONIC
{KGSAnalytics
,
212 title = {KGS Analytics
},
213 url
= {http
://kgs.gosquares.net
/},
217 @ELECTRONIC
{NaiveBayes1
,
218 author = {{K
}eselj
, {V
}lado and
{L
}in
, {Y
}ung
-chung and others
},
219 title = {{AI
::NaiveBayes1
} {CPAN
} Module
},
220 url
= {http
://search.cpan.org
/~vlado
/AI
-NaiveBayes1
-1.8/},
224 @ELECTRONIC
{RankComparison
,
225 author = {Sensei's Library
},
226 title = {Rank
--- worldwide comparison
},
227 url
= {http
://senseis.xmp.net
/?RankWorldwideComparison
},
231 @ELECTRONIC
{KohonenPy
,
233 title = {python
-kohonen
, {A
} library of
{Kohonen
} maps
},
234 howpublished = {Released under
{MIT
} License
},
235 url
= {http
://code.google.com
/p
/python
-kohonen
/},
237 timestamp
= {2010.03.10}
241 author = {{M
}oud\v
{r
}\'
{i
}k
, {J
}osef and
{B
}audi\v
{s
}, {P
}etr
},
242 title = {GoStyle
--- Determine playing style in the game of Go
},
243 url
= {http
://repo.or.cz
/w
/gostyle.git
},
247 @TECHREPORT
{Nissen2003
,
248 author = {S. Nissen
},
249 title = {Implementation of a Fast Artificial Neural Network Library
(fann
)},
250 institution = {Department of Computer Science University of Copenhagen
(DIKU
)},
252 note = {http
://fann.sf.net
},
254 timestamp
= {2010.03.10}
257 @ELECTRONIC
{Python25
,
258 author = {{Python Software Foundation
}},
261 title = {Python
2.5},
262 url
= {http
://www.python.org
/dev
/peps
/pep
-0356/},
264 timestamp
= {2009.04.29}
267 @INPROCEEDINGS
{Riedmiller1993
,
268 author = {Martin Riedmiller and Heinrich Braun
},
269 title = {{A Direct Adaptive Method for Faster Backpropagation Learning
: The
271 booktitle = {IEEE International Conference on Neural Networks
},
275 timestamp
= {2010.03.07}
279 author = {J. L. Rodgers and W. A. Nicewander
},
280 title = {{Thirteen ways to look at the correlation coefficient
}},
281 journal = {The American Statistician
},
290 @INPROCEEDINGS
{SpatPat
,
291 author = {Stern
, David and Herbrich
, Ralf and Graepel
, Thore
},
292 title = {Bayesian pattern ranking for move prediction in the game of Go
},
293 booktitle = {ICML '
06: Proceedings of the
23rd international conference on Machine
297 address = {New York
, NY
, USA
},
299 doi
= {http
://doi.acm.org
/10.1145/1143844.1143954},
300 isbn
= {1-59593-383-2},
301 location
= {Pittsburgh
, Pennsylvania
}
304 @INPROCEEDINGS
{TeamProf
,
305 author = {{S
}rb
, {T
}omas and
{S
}\'
{y
}kora
, Jiri and
{B
}ahbouh
, {R
}advan and
306 {H
}\"
{o
}schl
, {C
}yril
},
307 title = {Using Visualisation of Group Structure and Dynamics in Team Development
},
308 booktitle = {50th Conference of the International Military Training Association
},
310 location
= {Amsterdam
}
314 author = {Zito Tiziano and Wilbert Niko and Wiskott Laurenz and Berkes Pietro
},
315 title = {Modular toolkit for
{D
}ata
{P
}rocessing
({MDP
}): a
{P
}ython data
316 processing framework
},
317 journal = {Frontiers in Neuroinformatics
},
320 abstract = {Modular toolkit for Data Processing
(MDP
) is a data processing framework
321 written in Python. From the user
&#
39;s perspective
, MDP is a collection
322 of supervised and unsupervised learning algorithms and other data
323 processing units that can be combined into data processing sequences
324 and more complex feed
-forward network architectures. Computations
325 are performed efficiently in terms of speed and memory requirements.
326 From the scientific developer
&#
39;s perspective
, MDP is a modular
327 framework
, which can easily be expanded. The implementation of new
328 algorithms is easy and intuitive. The new implemented units are then
329 automatically integrated with the rest of the library. MDP has been
330 written in the context of theoretical research in neuroscience
, but
331 it has been designed to be helpful in any context where trainable
332 data processing algorithms are used. Its simplicity on the user
&#
39;s
333 side
, the variety of readily available algorithms
, and the reusability
334 of the implemented units make it also a useful educational tool.
},
335 doi
= {10.3389/neuro
.11/008.2008},
336 issn
= {ISSN
1662-5196},
338 timestamp
= {2010.03.05}
341 @comment
{jabref
-meta
: selector_publisher
:}
343 @comment
{jabref
-meta
: selector_author
:}
345 @comment
{jabref
-meta
: selector_journal
:}
347 @comment
{jabref
-meta
: selector_keywords
:}