• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to project page

2023 Fiscal Year Final Research Report

A Study on Adaptive Distributed Real-time Machine Learning Infrastructure

Research Project

  • PDF
Project/Area Number 19H04089
Research Category

Grant-in-Aid for Scientific Research (B)

Allocation TypeSingle-year Grants
Section一般
Review Section Basic Section 60050:Software-related
Research InstitutionNational Institute of Informatics

Principal Investigator

Takefusa Atsuko  国立情報学研究所, アーキテクチャ科学研究系, 教授 (70345411)

Co-Investigator(Kenkyū-buntansha) 小口 正人  お茶の水女子大学, 基幹研究院, 教授 (60328036)
中田 秀基  国立研究開発法人産業技術総合研究所, 情報・人間工学領域, 主任研究員 (80357631)
Project Period (FY) 2019-04-01 – 2022-03-31
Keywordsクラウドコンピューティング / エッジコンピューティング / リアルタイム処理 / 機械学習処理基盤 / IoT
Outline of Final Research Achievements

Cyber-physical systems (CPS) that can learn and infer from various sensor data in real-time and utilize them in various services and control systems are becoming increasingly important. In particular, there is a growing demand for analyzing video images, as a lot of information can be obtained from them, but the amount of calculations and data required for them is enormous. Therefore, it is necessary to effectively process them by combining edge and cloud computing resources.
In this research, we aimed to build a highly efficient, adaptive, wide-area, real-time machine learning processing platform for CPS that involves video analysis. We constructed an edge-cloud video distributed machine learning platform, developed an indoor video analysis method, and constructed a sensor-edge-cloud environment using a ROS-compliant robot application.

Free Research Field

計算機システム

Academic Significance and Societal Importance of the Research Achievements

センサ,エッジ,クラウドに分散した機械学習および推論処理を可能にする計算基盤は,Society 5.0で必要とされるサイバーフィジカルシステム(CPS)の構築技術として必須のものである.本研究で構築したエッジ・クラウド動画像分散機械学習基盤により得られた知見は,効率のよいCPSの構築に貢献することが期待できる.また,合成データを用いた動画像解析技術は,動画像を扱う多くのアプリケーションや動画像を対象とした生成AIへ適用可能である.

URL: 

Published: 2025-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi