T. Kojima. Automatic Acquisition of Go Knowledge from Game Records: Deductive and Evolutionary Approaches. PhD thesis, University of Tokyo, 1998. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
Computer Go - Müller (2000) (1 citation) (Correct)
....8 An extensive bibliography of online papers about computer Go by Markus Enzenberger is currently located at home:t online:de=home=markus:enzenberger =compgo biblio:html. Ph.D. theses about computer Go started appearing 30 years ago, and are recently published at a rate of about one per year [6, 12, 22, 24, 28, 32, 37, 43, 57, 59, 60, 62, 86]. 2.3 Go Research in Related Fields Go has been used as the topic for research in related elds such as cognitive science, software engineering and machine learning [11, 8, 12, 18, 29, 33, 59, 61, 62, 63, 85, 74] Starting from only the rules of the game, learning programs can pick up basic Go ....
....to endgame. Almost all Go programs contain a pattern database and a pattern 11 matching subsystem. Most pattern databases are still generated by hand, and contain several thousand patterns. However, methods have been developed that can automatically extract patterns from professional games [84, 32, 33, 34, 65, 67], and such a pattern learning system is used in at least one of the top programs. Many variants of pattern matching problems have been studied in computer science. In Go, a large number of 2 dimensional patterns must be compared to a single full board position in the 16 di erent ways illustrated ....
T. Kojima. Automatic Acquisition of Go Knowledge from Game Records: Deductive and Evolutionary Approaches. PhD thesis, University of Tokyo, 1998.
Knowledge Acquisition from Game Records - Kojima, Yoshikawa (1999) Self-citation (Kojima) (Correct)
....be interpreted by the forcing rules of the system is the move to be learned (Step 2) 3 4 5 1 2 f1:defense f1:attack learning position learning move interpretation Figure 5: Finding a move to be learned. A reason why the move should be learned is briefly explained (detailed explanation is in [ 8 ] ) Since the system is beaten, the last move (Move 5 in Figure 5) is interpreted as attack by a winner and the move before the last (Move 4) is interpreted as defense by the system. Other moves are also interpreted alternately attack by a winner and defense by the system. The last move which ....
....algorithm acquires a proper rule as shown in Figure 11. yyxx y xy 5 . x 1 3 y Figure 10: A problem and its answer of Geta. yy yi V V y V P V Figure 11: An acquired rule of Geta. This system automatically detects inaccurate rules and refines them, which is not explained in this paper. See [ 8 ] . 4 AN EVOLUTIONARY APPROACH This approach is taken to acquire a large amount of heuristic knowledge from a large amount of training examples. 4.1 ALGORITHM In this section, the proposed algorithms are explained. See [ 8, 9, 10 ] for details. 4.1.1 Overview of the Algorithm The algorithm s ....
[Article contains additional citation context not shown here]
T. Kojima. Automatic Acquisition of Go Knowledge from Game Records: Deductive and Evolutionary Approaches. PhD thesis, The University of Tokyo, 1998.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC