A New Model of Reflective Introspective Learning

May 1998

Proceedings of the Eleventh International Florida Artificial Intelligence Research Society Conference (FLAIRS 1998)

Volume

Proceedings of the Eleventh International Florida Artificial Intelligence Research Society Conference (FLAIRS 1998)

All Papers

Systems which introspect about their own processes can improve their reasoning behavior in response to experience using "introspective learning" techniques. Many systems which perform introspective learning an-alyze and change only an underlying domain task’s reasoning processes. They do not possess the ability to reflectively introspect about the introspective task itself. We present a model of a truly reflective introspective learning system which uses the same case-based reason-ing mechanisms for both its domain and introspective tasks. The reuse of case-based reasoning mechanisms enables the reuse of introspective models and mechanisms developed for a CBR domain reasoner to reflect about the introspective process itself. We suggest an introspective model capable of reflection, and begin to address the issue of balancing introspection and action within a reflective system.