Data-Driven Learning Optimization of Dynamical System with Stochastic Uncertainty and Its Application

2022 Fiscal Year Final Research Report

• Project/Area Number: 19K15019
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 21040:Control and system engineering-related
• Research Institution: Tokyo Metropolitan University
• Principal Investigator: Toyoda Mitsuru 東京都立大学, システムデザイン研究科, 助教 (40826939)
• Project Period (FY): 2019-04-01 – 2023-03-31
• Keywords: 制御理論 / 最適制御 / 確率論理システム / スパース最適化

Outline of Final Research Achievements

This research addressed control and optimization problems related to the analysis of dynamical systems and focused on the development of new analysis framework based on the data science methodologies. In probabilistic Boolean networks, which is a class of discrete-valued systems taking into consideration the stochastic uncertainty, optimization algorithms in control and estimation problems were presented. Furthermore, optimization problems with the l1 norms for the regularization were considered in this study. Based on the sparse optimization method in statistic and machine learning re-search fields, multiple iterative methods were examined, and associated convergence analysis was performed.

Free Research Field

制御工学

Academic Significance and Societal Importance of the Research Achievements

離散値システムの表現として本研究で主に用いられた確率ブーリアンネットワークは汎用的な数学モデルであり，本研究で提案した最適化手法をはじめとする解析結果は広いクラスのシステムの研究に関連するものである．また，本研究における連続値システムの解析で主として扱ったスパース最適化手法もまた制御，最適化，機械学習など多くの分野で取り入れられている手法であり，こちらも他分野への適用といった更なる発展が期待される．