Integrated Learning with Incorrect and Incomplete Theories

Abstract This paper discusses incorrect and incomplete theories in integrating empirical and explanation-based learning techniques. The paper focuses on OCCAM, a program which is unique among explanation-based learning systems in that it has the capability to acquire via empirical means the knowledge needed for explanation-based learning. Two major extensions to OCCAM are reported: • The ability to revise a schema formed by explanation-based learning when it becomes apparent that the underlying domain theory was incorrect. • The ability to use empirical learning techniques to acquire a new rule to complete an explanation. These extensions address the issue of utilizing incorrect and incomplete theories in an integrated learning system.