基于归纳的机器学习方法的实现研究

On Implementing a Method of Inductive-Based Machine Learning

一、引言机器学习是解决计算机获取知识问题的方法。在这项研究中,对人类学习方法的模拟是一条重要的途径,而归纳正是人类学习的重要方式之一。从归纳中获取知识是人类新知识的极其重要的来源,它通过对经验材料的研究,探求新知识。由于一切科学理论的基本前提最初都是从经验事实中总结出来的,而从经验到一般规律的过程中,归纳推理起着不可替代的作用。

This paper analyzes the method of inductive learning and emphasizes the importance of knowledge during the inductive processes.From this view, we devise an inductive-based machine learning model LEARNER/TH-1 and an inductive algorithm constructing step by step.In this model,the system uses the domain knowledge on a large scale,and the inducted cases are looked upon as a kind of experienced knowledge.The background knowledge, the experienced knowledge and the inductive conclusion have the same repres- entation.form.The learning process is self-drive.