P. Smyth, The Application of Information Theory to Problems in Decision Tree Design and Rule-Based Expert Systems, PhD thesis, California Institute of Technology, 1988.  Home/Search   Document Not in Database   Summary   Related Articles

....its domain Unfortunately, while the above algorithms are fascinating from a cognitive science viewpoint, the symbolic AI approach tends to break down when given noisy examples and fails to scale up to problems of larger dimensionality. 1 For a more detailed history of rule based systems, see [Smy88] 8 More recently, interest in database analysis has generated methods for discovery of rules. Gaines and Shaw [GS86] use fuzzy logic to induce inference rules about a very specific type of database (a repertory grid) Piatetsky Shapiro [Pia89] presents a method for learning strong rules from ....

....measure of the value of a rule. This allows us to objectively state which of two candidate rules is the better. 3.1. 2 The ITRULE Algorithm Now that we can rank rules, how can we search the space of all possible rules to find the best The Information Theoretic RULE induction algorithm (ITRULE) Smy88] was 12 developed for just this purpose. 1 The user will specify two parameters to limit our search: first, the maximum order (number of conditions) d beyond which we will not search, and second, the number of rules R he wishes to obtain. Since the best rules are more likely to be lower in ....

[Article contains additional citation context not shown here]

P. Smyth, The Application of Information Theory to Problems in Decision Tree Design and Rule-Based Expert Systems, PhD thesis, California Institute of Technology, 1988.

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