Press "Enter" to skip to content

Get A Theory of Heuristic Information in Game-Tree Search PDF

By Chun-Hung Tzeng

ISBN-10: 3642613683

ISBN-13: 9783642613685

ISBN-10: 3642648126

ISBN-13: 9783642648120

Searching is a vital approach in so much AI structures, specially in these AI creation platforms inclusive of a world database, a collection of creation principles, and a keep an eye on method. as a result of the intractability of uninformed seek methods, using heuristic info is important in such a lot looking techniques of AI structures. this crucial inspiration of heuristic informatioD is the imperative subject of this ebook. We first use the 8-puzzle and the sport tic-tac-toe (noughts and crosses) as examples to aid our dialogue. The 8-puzzle includes 8 numbered movable tiles set in a three x three body. One mobilephone of the body is empty in order that it's attainable to maneuver an adjoining numbered tile into the empty phone. Given tile configurations, preliminary and target, an 8-puzzle challenge comprises altering the preliminary configuration into the objective configuration, as illustrated in Fig. 1.1. an answer to this challenge is a chain of strikes top from the preliminary configuration to the aim configuration, and an optimum resolution is an answer having the smallest variety of strikes. no longer all difficulties have suggestions; for instance, in Fig. 1.1, challenge 1 has many strategies whereas challenge 2 has no resolution at all.

Show description

Read or Download A Theory of Heuristic Information in Game-Tree Search PDF

Best theory books

Download e-book for iPad: Angle and Spin Resolved Auger Emission: Theory and by Bernd Lohmann

The Auger impact has to be interpreted because the radiationless counterpart of photoionization and is generally defined inside a two-step version. perspective and spin resolved Auger emission physics offers with the theoretical and numerical description, research and interpretation of such kinds of experiments on loose atoms and molecules.

Radu Precup (auth.)'s Methods in Nonlinear Integral Equations PDF

Tools in Nonlinear fundamental Equations provides a number of tremendous fruitful equipment for the research of structures and nonlinear critical equations. They comprise: mounted aspect tools (the Schauder and Leray-Schauder principles), variational equipment (direct variational equipment and mountain move theorems), and iterative tools (the discrete continuation precept, top and decrease options suggestions, Newton's process and the generalized quasilinearization method).

Extra info for A Theory of Heuristic Information in Game-Tree Search

Sample text

For even h( > 1), let N z (2 denotes the height of the node) be the node of the game position consisting of the middle 3 squares of the playing board of N. The playing board of N z is given by removing (h/2-1) squares from each end of N; N z is the middle descendant of N (if N z is not N itself) of height 2. Similarly, if h (> 1) is odd, then let N 1 (of height 1) be the middle descendant of N, given by removing (h-l)/2 squares from each end of N. Then we have the following theorem. 1. " Proof Suppose that h > 4 is even; that is, N is a MAX node.

Not a WIN) node if and only if all of its sons are LOSS nodes. Therefore, the conditional probability that A is a LOSS node is I-p=n (I-Pi) i if all of its sons are independent. The probability P of a WIN at A is (I) p=l-n (l-pJ. i Case 2: A is a MIN node. Then A is a WIN node if and only if all of its sons are WIN nodes. Therefore, the probability p of A being a WIN node is (II) if all of its sons are independent. 24 Heuristic Game-Tree Searches The rules (I) and (II) are called the product-propagation rules, and the corresponding back-up process is called the product-propagation procedure, first proposed by Pearl (1981).

For example, the Pz-game model for the game tree in Fig. 4 is a probabilistic model, in which the space {O, 1}16 is the sample space with the probability measure induced by the probabilities of 0 and 1 at each terminal node. In this model there are a total of 2 16 = 65536 possible games. The G I-game model for the game graph in Fig. 2 is another probabilistic game model, in which the sample space is {O, 1}5. There are only 2 5 = 32 potential games in this model. 2 Strategies and Game Values Let (Q, F, P) be a probabilistic game model for a game graph T, and let X be a Tgame assigned to the game graph T.

Download PDF sample

A Theory of Heuristic Information in Game-Tree Search by Chun-Hung Tzeng

by Christopher

Rated 4.73 of 5 – based on 17 votes