Learning Based on Linguistic Instructions Using Fuzzy Theory(Journal of Japan Society for Fuzzy Theory and Systems)

概要

論文の詳細を見る
Human learning using language is superior to other kind of learning. Human learning on a linguistic level is not a recursive learning by random trial and error but the acquisition of knowledge by combining linguistic instructions with intrinsic knowledge. In this paper, we propose a framework of learning based on linguistic instructions using fuzzy theory FULLINS(FUzzy-Learning based on Linguistic INStruction). In building learning systems that acquire knowledge from linguistic instructions, the validity of learning depends on how to determine certain functional requirements. Intrinsic ability that understands linguistic instructions is required in such systems. We aim to build a learning system called FULLINS, where the performance of the system is improved by a supervisor's linguistic instructions. Hence, FULLINSis a mechanism for learning of both concepts and language. FULLINS consists of six functional components : Task Performance, Dialogue, Explanation, Background Knowledge, Interpretation, and Self-Regulating. The validity of FULLINS is verified by applying it to two examples, a control system, The Truck Backer-Upper problem, and a board game, Five-in-a-Row.
日本知能情報ファジィ学会の論文
1992-12-15