Validation Method for Intelligent Systems

Setuo Tsuruta, Hitachi Ltd., Japan; Takashi Onoyama, Sen Kubota and Kazuko Oyanagi, Hitachi Software Engineering Co., Ltd., Japan

Knowledge incorporated intelligent solving methods are prevailing in practical planning and scheduling because of the large problem size and complex constraints. However, when intelligent methods are used in real applications, it is an important problem to validate that both optimality and response time are at expert level. Turing Test approaches are useful in validating intelligent systems. Nevertheless, validation burdens experts with heavy tasks. In distribution route scheduling, for instance, validation load is considerable since patterns of distribution spots are numerous. A bi-directional, many-sided explanation typed multi-step validation method, which enables to share the validation tasks among experts, KEs and computers, is proposed to diminish the load on busy experts, for validating intelligent systems.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.