Machines that Learn to Play Games

Front Cover
Nova Publishers, Jan 1, 2001 - Games - 298 pages
1 Review
The mind-set that has dominated the history of computer game playing relies on straightforward exploitation of the available computing power. The fact that a machine can explore millions of variations sooner than the sluggish human can wink an eye has inspired hopes that the mystery of intelligence can be cracked, or at least side-stepped, by sheer force. Decades of the steadily growing strength of computer programs have attested to the soundness of this approach. It is clear that deeper understanding can cut the amount of necessary calculations by orders of magnitude. The papers collected in this volume describe how to instill learning skills in game playing machines. The reader is asked to keep in mind that this is not just about games -- the possibility that the discussed techniques will be used in control systems and in decision support always looms in the background.
 

What people are saying - Write a review

We haven't found any reviews in the usual places.

Contents

631 Experimental details
121
632 Simple 1ply training
122
633 Training with 1ply search plus expansions
123
64 Tuning Deep Blues evaluation function
125
641 Modified training algorithm
126
642 Effect on the KasparovDeep Blue rematch
127
65 Discussion
129
Feature Construction for Game Playing
131

223 Learning from mistakes
17
224 Learning from simulation
19
24 Evaluation Function Tuning
21
241 Supervised learning
22
242 Comparison training
24
243 Reinforcement learning
28
244 Temporaldifference learning
30
245 Issues for evaluation function learning
33
25 Learning Patterns and Plans
41
251 Advicetaking
42
252 Cognitive models
43
253 Patternbased learning systems
46
254 Explanationbased learning
48
255 Pattern induction
50
256 Learning playing strategies
53
26 Opponent Modeling
55
27 Conclusion
58
Human Learning in Game Playing
61
32 Research on memory and perception
63
321 Memory
64
322 Perception
67
The chunking and template theories
68
33 Research on problem solving
70
331 De Groots results
71
332 Theories and computer models of problem solving
72
34 Empirical studies of learning
75
341 Shortrange learning
76
342 Mediumrange learning
77
343 Longrange learning
78
35 Human and machine learning
79
36 Conclusions
80
Toward Opening Book Learning
81
42 Basic Requirements
82
43 Choosing Book Moves
83
44 Book Extension
84
45 Implementation Aspects
86
46 Discussion and Enhancements
87
47 Outlook
88
Reinforcement Learning and Chess
91
52 KnightCap
94
522 Search algorithm
95
524 Search extensions
96
527 Move ordering
97
5210 Modification for TDLeaf𝛌
98
53 The TD𝛌 algorithm applied to games
100
54 Minimax Search and TD𝛌
102
55 TDLeaf𝛌 and Chess
105
552 Discussion
111
56 Experiment with Backgammon
113
562 Experiment with LGammon
114
57 Future Work
115
Comparison Training of Chess Evaluation Functions
117
62 Comparison Training for ArbitraryDepth Searches
119
63 Tuning the SCP evaluation function
120
72 Evaluation Functions
132
73 Feature Overlap
133
74 Constructing Overlapping Features
134
742 Higher order expansion
136
743 Quasirandom methods
138
75 Directions for Constructing Overlapping Features
139
752 Compression
142
753 Teachable systems
143
76 Discussion
151
Learning to Play Expertly A Tutorial on Hoyle
153
82 A GamePlaying Vocabulary
154
83 Underlying Principles
156
831 Useful knowledge
157
832 The Advisors
159
833 The architecture
160
834 Weight learning for voting
164
84 Perceptual Enhancement
166
841 Patterns
167
842 Zones
168
85 An Empirical Framework
172
86 Results
173
Why Hoyle Works
176
Acquisition of Go Knowledge from Game Records
179
912 Classification of Go Knowledge
180
913 Two Approaches
181
93 A Deductive Approach
183
932 Rule Acquisition
186
94 An Evolutionary Approach
190
941 Algorithm
191
942 Application to TsumeGo
196
95 Conclusions
202
Honte a GoPlaying Program Using Neural Nets
205
1011 Rules
206
1012 Strength of programs
207
102 General Approach in Honte
210
103 Joseki Library
212
105 AlphaBeta Search
215
106 Influence
218
108 Evaluation of Honte
220
109 Conclusions
223
Learning to Play Strong Poker
225
112 Texas Holdem
228
113 Requirements for a WorldClass Poker Player
229
114 Lokis Architecture
231
115 Implicit Learning
233
116 Explicit Learning
236
117 Experiments
238
118 Ongoing Research
240
119 Conclusions
242
Bibliography
243
Contributors
269
Person Index
275
Subject Index
281
Copyright

Common terms and phrases

Popular passages

Page iv - ISBN 1-56072-865-5 Volume 4 - Structured Matrices: Recent Advances and Applications Dario Andrea Bini, Eugene Tyrtyshnikov and Plamen Yalamov ISBN 1-56072-890-6 Volume 3 - Recent Trends in Numerical Analysis Donate Trigiante ISBN 1-56072-885-X Volume 2 - High Performance Algorithms for Structured Matrix Problems Peter Arbenz, Marcin Paprzycki, Ahmed Sameh and Vivek Sarin ISBN 1-56072-594-X Lee Keener, Editor-in-chief College of Science and Management, University of Northern British Columbia Prince...
Page 248 - Experiments with Multi-ProbCut and a new high-quality evaluation function for Othello. In HJ van den Herik and H.
Page 268 - H. Liu and H. Motoda (Eds.) Feature Extraction, Construction and Selection: A Data Mining Perspective.

Bibliographic information