A new method for explanation-based learning

摘要

In this paper we discuss using the stratified ATMS to realize explanation-based learning. As the stratified ATMS can record and maintain the reasonings for beliefs efficiently and can deal with non-monotonic reasoning, so the ATMS-based EBL system can improve the efficiency of explanation-based learning, deal with multiple explanation problems in learning from imperfect theories by prioritized reasoning and multiple example verification and can give biases for induction in integrated learning.