Examples' Depth and Induction of Recursive Logic Programs
スポンサーリンク
概要
- 論文の詳細を見る
The main goal of a Inductive Logic Programming (ILP) system is to induce a logic program that explains a given set of positive examples and is consistent with negatives examples. Inductive Logic Programming inherist concepts and methods from Machine Learning and Logic Programming. An important class of logic programs is the class of recursive programs. Many programs for list processing and mathematical functions are expressed using recursive logic programs. Recursive programs consist of at least two clauses, which are named base and recursive clauses. A recursive clause has one or more literals in its body with the same predicate of clause's head. It has not been possible to induce correct recursive programs, using existing ILP systems, without carefully selected set of training examples. An analysis of these training sets shows that the depth of an examples plays an important role on induction of recursive programs : if examples of any depth k aren't present in the given set of examples, induction can't be performed correctly. The main reason for this problem is that existing ILP systems use induction operators based on θ-subsumption instead of operators based on implication. Operators based on θ-subsumption are incomplete with respect to implication. In order to overcome this problem, we use an approach called forced simulation. Basically, the forced simulation algorithm simulates (in a controlled way) the hypothesized recursive program for each example. This is a simple technique to invert implication
- 一般社団法人情報処理学会の論文
- 1995-09-20
著者
-
石塚 満
Dept. Of Information And Communication Engineering Faculty Of Engineering The University Of Tokyo
-
Gomi Satoshi
Dept. of Information and Communication Engineering, Faculty of Engineering, The University of Tokyo
-
Gomi Satoshi
東京大学工学部電子情報工学科
関連論文
- Examples' Depth and Induction of Recursive Logic Programs
- Induction of 2-clause Recursive Logic Programs Using Minimal Multiple Generalization and Forced Simulation