Geoffrey I. Webb. Techniques for efficient empirical induction. In C. J. Barter and M. J. Brooks (editors), AI'88 -- Proceedings of the Third Australian Joint Conference on Artificial Intelligence, pages 225--239, Adelaide, 1990. Springer-Verlag.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....search strategies for unordered search The OPUS o search algorithm [18] has demonstrated the capacity to efficiently search many common classification learning search spaces. Prior to OPUS o , fixed order search was used exclusively for admissible search in classification learning [14, 15, 16, 17]. Fixed order search orders the search space in advance so that the proportion of the search space under each node is predetermined. In contrast, OPUS o can reorder the search space to optimize the proportion of the search space under each node. OPUS o outperforms fixed order search by ....

Geoffrey I. Webb. Techniques for efficient empirical induction. In C. J. Barter and M. J. Brooks (editors), AI'88 -- Proceedings of the Third Australian Joint Conference on Artificial Intelligence, pages 225--239, Adelaide, 1990. Springer-Verlag.

....of the search space that may be pruned. Rather, it takes as input functions for identifying sections of the search space for pruning, and ensures that maximal advantage is obtained from each pruning action. A number of recent machine learning algorithms have performed restricted systematic search [6 8]. All of these algorithms have been based on an organisation of the search space, that, when considering the search problem illustrated in Figures 1 to 3, traverse the search space in the manner depicted in Figure 4. Such an organisation is achieved by arranging the operators in a predefined ....

....b. Set n .search map to REMAINING OPERATORS. c. Add n to OPEN. 8. Go to step 2. OPUS s ensures that no state is examined more than once (unless identical states can be formed by different combinations of operator applications) using a similar search space organisation strategy to that of Webb [6], Rymon [8] and Schlimmer [7] It differs, however, in that instead of placing the largest subsection of the search space under the highest ordered operator, the second largest subsection under the second highest ordered operator, and so on, whenever pruning occurs, the largest possible proportion ....

Webb, G.I., Techniques for efficient empirical induction, in AI'88, C.J. Barter and M.J. Brooks, Editor^Editors. 1990, Springer-Verlag: Berlin. p. 225-239.

....Systematic search strategies for unordered search The OPUS o search algorithm [15] has demonstrated the capacity to efficiently search many common classification learning search spaces. 4 Prior to OPUS o , fixed order search was used exclusively for systematic search in classification learning [11 14]. Fixed order search orders the search space in advance so that the proportion of the search space under each state is predetermined. In contrast, OPUS o can reorder the search space to optimize the proportion of the search space under each state. OPUS o outperforms fixed order search by ....

....search space below a state) as does exclusive pruning. While inclusive pruning has been demonstrated in the context of the OPUS o search algorithm, it could also be applied to fixed operator ranking algorithms for unordered search such as those of Rymon [11] Schlimmer [12] Segal [13] and Webb [14]. Inclusive pruning axioms for search in classification learning have been defined and demonstrated to deliver substantial savings in the proportion of the search space traversed (in the case of the Wisconsin Breast Cancer data, a reduction by 61 ) Inclusive pruning is by no means, however, ....

Webb, G.I., Techniques for efficient empirical induction, in AI'88, C.J. Barter and M.J. Brooks, Editors. 1990, Springer-Verlag: Berlin. p. 225-239.

No context found.

Webb, G. (1990) Techniques for efficient empirical induction. In C. J. Barter & M. J. Brooks (Eds) AI `88. Springer-Verlag, Berlin, pp. 225--239.

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