Creation of the acceptable intelligent system based on the system intelligence circulation

2021 Fiscal Year Final Research Report

• Project/Area Number: 19H02159
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Single-year Grants
• Section: 一般
• Review Section: Basic Section 21040:Control and system engineering-related
• Research Institution: Yokohama National University
• Principal Investigator: HAMAGAMI TOMOKI 横浜国立大学, 大学院工学研究院, 教授 (30334204)
• Project Period (FY): 2019-04-01 – 2022-03-31
• Keywords: 知能システム / 機械学習 / 人工知能

Outline of Final Research Achievements

These studies have achieved four main results toward the circulation mechanism of system knowledge for acceptable intelligent systems and system creation. First, we realized “adaptive threshold boosting” as a method for distilling knowledge extraction from experts and separating common and specific knowledge. Second, we clarified a method for adaptive partitioning and transferring multi-domain spaces by multi-agent approach for the advancement of machine learning valuation models in multi-domain spaces. Third, machine learning provides visual feedback and improved explanatory power to the shape design and refinement process. Furthermore, we realized the control problem by hierarchical reinforcement learning, and reused and circularized the intellectual structure by extracting its basic components.

Free Research Field

知能システム

Academic Significance and Societal Importance of the Research Achievements

サイバーフィジカルシステム(Cyber Physical Systems: CPS)は，今後の社会システムの姿を大きく変えるパラダイムである。CPSと人工知能の融合のためには，知能システム自らが，システム知を分析・説明・制御する仕組み持ち，この仕組みの循環を通して，XAIを超えた受容可能な知能システム(Acceptable Intelligent Systems: AIS)へと発展することが必要である。本研究はそのために必要なシステム知の抽出と説明法，設計と実装法，応用方法について，要素技術から具体的アプリケーションまで実現したことで，本分野の学術的および社会的に大きな意義を有する。