Hierarchical decision-making techniques using Bayesian estimation for IoT systems

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K11845
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 60060:Information network-related
• Research Institution: Osaka University
• Principal Investigator: Daichi Kominami 大阪大学, 大学院情報科学研究科, 助教 (00709678)
• Project Period (FY): 2021-04-01 – 2024-03-31
• Keywords: ベイズ推論 / 意思決定 / 無線ネットワーク / 集団的意思決定

Outline of Final Research Achievements

In recent years, mathematical models have been proposed to explain the efficient information processing performed by the human brain. The Bayesian attractor model, which imitates the top-down decision-making process by the brain using Bayesian estimation, is one such model. In this research project, I proposed methods to aggregate and integrate the decision-making information from multiple nodes when they make decisions according to the Bayesian attractor model. This integrated decision-making model is based on the Bayesian causal inference. This decision-making model integrates the posterior probability density, which expresses which of multiple alternatives is most plausible, based on the reliability of the observed values, and we confirmed that it can make adaptive decisions to noise in the observed values.

Free Research Field

情報ネットワーク

Academic Significance and Societal Importance of the Research Achievements

無線通信技術の発展やIoTの普及により、情報観測を行う機能を備えた多数の機器がネットワークを介して相互に接続するようになっている。機器からの観測情報を活用する際には、観測情報に基づき適切な意思決定を行うことが重要となる。本研究課題では、観測情報の種類ごとに信頼性を計算し、信頼できる観測情報を重視して統合する意思決定手法を提案した。これにより、一部の観測値の誤りに耐性を備えた意思決定手法を構築した。