Registered Data

[00819] Secure Computing: Maintaining Personal Privacy while Analyzing Data

  • Session Time & Room : 3D (Aug.23, 15:30-17:10) @E804
  • Type : Proposal of Industrial Minisymposium
  • Abstract : The proliferation of IoT machine and sensor devices are major contributors to the surge in production and storage of mostly private, sensitive data. Deloitte projects, "by 2025 our global volume will reach 175 zeta bytes" (www2.deloitte.com/cy/en/pages/technology/articles/data-grown-big-value.html). Many data owners seek to analyze the data to uncover insights and improve their decision-making processes. However, compliance with privacy regulations and the threat of cyberattacks pose heretofore unknown challenges. Approaches to address this issue, collectively known as secure computing, include: privacy-preserving data analysis, differential privacy, federated learning, multi-party computation, and homomorphic encryption. This mini-symposium seeks to gather practitioners/specialists for active debate and dialog.
  • Organizer(s) : Mei Kobayashi
  • Classification : 68M25, Computer security, Privacy of data, Data encryption (aspects in computer science), Artificial neural networks and deep learning
  • Minisymposium Program :
    • 00819 (1/1) : 3D @E804 [Chair: Mei Kobayashi]
      • [04041] Construction of Differentially Private Summary with Homomorphic Encryption
        • Format : Talk at Waseda University
        • Author(s) :
          • HAYATO YAMANA (Waseda University)
          • Shojiro USHIYAMA (Waseda University)
          • Tsubasa TAKAHASHI (LINE Corp.)
        • Abstract : A differentially private summary for range queries is constructed using homomorphic encryption to hide the raw data from the computation server. To shorten the processing time, we proposed a new method to merge adjacent close values in the histogram if the difference between the adjacent data is small. Then, we confirmed that the accuracy of the proposed method was equivalent to a state-of-the-art algorithm and the processing time is O(n).
      • [04961] Federated Learning with Differential Privacy and Secure Computing
        • Format : Talk at Waseda University
        • Author(s) :
          • Tsubasa Takahashi (LINE Corporation)
        • Abstract : Improving user experience while respecting user privacy is important nowadays. Last year we released federated learning in LINE messenger’s keyboard area to make users sticker selection easier and more personalized while preserving user privacy. Our FL also employs Differential Privacy (DP) to make exploiting user privacy more difficult. This talk presents our FL+DP, and recent advancement of privacy amplification with secure computing.
      • [05348] Network Security and Analytics for Reliability
        • Format : Talk at Waseda University
        • Author(s) :
          • Yukio Uematsu (Tokyo University of Science/Nokia)
        • Abstract : In recent years, a vast number of IoT devices have been connected to the cloud through mobile networks. This talk will address two fundamental issues, namely security and data reliability, in the context of mobile network data analytics. We will present some use cases for data management that combine edge and cloud computing while ensuring data reliability.