Integrated data analysis on multiple institutions 分散データ非共有統合解析技術





  • 複数機関が分散保持するデータの生データを共有せずに統合解析するデータコラボレーション解析技術の開発
    • データコラボレーション解析技術は、各企業・機関が保有するプライバシー情報などを含む元データの代わりに、元データをAI技術により変換した「中間表現データ」のみを共有し、複数機関のデータの統合解析を可能とします。これにより、元データに含まれる秘匿性の高い情報の安全性を担保しつつ、多数のデータの取り扱いが可能となることでAIの解析精度の大幅な向上を実現します。
    • 研究テーマ例
      • 各種のデータ・タスクに対応するデータコラボレーション解析技術の開発
      • データコラボレーション解析技術の高度化
  • 企業や自治体、病院等の実データを用いたデータコラボレーション技術の応用
    • 共同研究を実施している企業・自治体、病院等の実データを利用したデータコラボレーション解析技術の実証実験を進めています。
    • 研究テーマ例
      • 実データを利用したデータコラボレーション解析技術の性能評価
      • 実データに合わせたデータ解析・前処理技術の開発



In recent years, as data collection and accumulation have become easier, various companies and institutions have been accumulating their own data and working on analysis using artificial intelligence (AI). To improve the performance of AI analysis, it is necessary to collect a sufficient amount of data. Therefore, it is expected that highly accurate analysis will be possible if data held by multiple institutions can be integrated for analysis. However, it is difficult to share the original data containing personal information and trade secrets across institutions. Therefore, there is a need for technology that enables integrated analysis across corporate and institutional boundaries without sharing the original data.


Outline and research topics

  • Development of the data collaboration (DC) analysis for integrated analysis of data held by multiple institutions without sharing.
    • Data collaboration (DC) analysis is a recent technology which enables integrated analysis of data from multiple institutions by sharing only “intermediate representation ” converted from the original data using AI technology, instead of the original data including privacy information. DC analysis enables the handling of a large amount of data while ensuring the security of highly confidential information contained in the original data, thereby significantly improving the performance of AI analysis.
    • Example of research topics:
      • Development of DC technology for various types of data and tasks.
      • Advancement of DC technology.
  • Application of DC technology for real data from companies, municipalities, hospitals, etc.
    • We are conducting demonstration tests of DC analysis for actual data from companies, municipalities, hospitals, etc., with which we are conducting joint research.
    • Example of research topics:
      • Performance evaluation of DC analysis for real-world data.
      • Development of data analysis and preprocessing techniques for actual data.