本文分析了当今机器学习现状,总结出三种学习体制:联结学习、遗传算法和符号学习。研究了三种体制各自的内涵和在智能活动中的地位及其相互间的联系。在此基础上,提出了综合不同学习方法的集成体制方案,并实现了一个基于解释的学习和基于相似学习的集成规则学习系统IR1。还讨论了联系学习与其它体制的局部集成。 This paper analyzes the current status of machine learning, summarizes three learning systems: learning, genetic algorithms and symbolic learning. Studied the connotation of each of the three systems and their status in the intelligent activities and their mutual relations. On this basis, we put forward an integrated system scheme that integrates different learning methods and implement an integrated learning system IR1 based on interpretation and similar learning. It also discusses the local integration of contact learning with other systems.