正例学习的一个启发式算法  

A Heuristic Algorithm for Learning from Positive Examples

作　　者：赵美德[1] 洪家荣[1] 王丽薇[1] 李星源 

机构地区：[1]哈尔滨工业大学计算机科学与工程系

出　　处：《电子学报》1994年第8期17-23,共7页Acta Electronica Sinica

摘　　要：本文基于多值逻辑函数极小化提出一种正例学习问题，并对这一正例学习问题给出一个启发式学习算法。同现存的示例学习方法相比，本文算法是基于演绎推理的，可对大量信息进行保真压缩。该算法具有简单、快速和易于实现的特点。本文用两个实例─焊接工艺自动设计知识获取和布尔函数极小化，将本算法同示例式归纳学习系统ＡＱ１５和布尔函数极小化系统ＳＷＴ进行了比较。This paper presents a learning problem from positive examples based on multiple valued minimization paradigm.A new heuristic algorithm for the problem is given.Compared with the existing learning methods from examples,the proposed algorithm is based on deductive reasoning. Furthermore,the algorithm is simple,fast and easy to implement. The comparison of the proposed algorithm with the inductive learning system AQ15 and the Boolean function minimization system SWT is carried out by two illustrations: the knowledge acquisition for the automatic weiding technology design and the Boolean function minimization.

关 键 词：多值逻辑函数 正例学习 启发式算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]