Challenges and Opportunities of Privacy-Preserving Machine Learning

  • Type:Bachelor, Master
  • Date:sofort/immediately
  • Supervisor:

    David Jin

  • Background:

    Machine learning exploded in popularity over the last decade but often requires access to sensitive data. Privacy restrictions prevent or limit the collection and processing of such sensitive data which is often seen as a challenge. 

    Privacy-preserving machine learning (PPML) approaches try to address these problems. Federated learning for example can be used to train machine learning models on otherwise inaccessible data. Unfortunately, these approaches require additional effort to implement and usually decrease performance. Such technical drawbacks of PPML are well understood, but other challenges or potential opportunities which arise when using such privacy-preserving techniques remain unknown. What are the challenges and opportunities of PPML in an organizational context?

     

    Objectives:

    Expert Interviews to identify and discuss challenges and opportunities of using PPML

     

    Introductory Literature:

    Liu, Bo, Ming Ding, Sina Shaham, Wenny Rahayu, Farhad Farokhi, and Zihuai Lin. “When Machine Learning Meets Privacy: A Survey and Outlook.” ACM Computing Surveys 54, no. 2 (March 5, 2021): 31:1-31:36. https://doi.org/10.1145/3436755.

    Casadesus-Masanell, Ramon, and Andres Hervas-Drane. “Competing with Privacy.” Management Science 61, no. 1 (January 2015): 229–46. https://doi.org/10.1287/mnsc.2014.2023.

    Harborth, David, Maren Braun, Akos Grosz, Sebastian Pape, and Kai Rannenberg. “Anreize und Hemmnisse für die Implementierung von Privacy-Enhancing Technologies im Unternehmenskontext,” 2018. https://doi.org/10.18420/SICHERHEIT2018_02.