一种基于扩张矩阵理论的规则学习算法

A NEW RULE LEARNING ALGORITHM BASED ON EXTENSION MATRIX THEORY

本文给出的扩张矩阵合并问题是ＮＰ困难问题的一种新的证明。由此得到了析取式最小问题是ＮＰ困难问题的新的证明。之后给出了两个扩张矩阵合并的完备策略，在此基础上给出了扩张矩阵合并的启发函数策略。

This paper gives a new kind of poof that to compound extension matrix is NP--hard. From this, we get a new poof that the minimum problem of disjunctiving formula is NP--hard. This paper gives two complete tactics of compounding extension matrix, so as heuristic function tactic that compound the extension matrix. At last the paper gives a new algorithm of the learning from examples-the heuristic function learning algorithm based on extension matrix theory.