Secure Multi-Party Computation and Data Sharing

Symposium
dcypher Symposium 2019 connecting cybersecurity knowledge – enterprises - policies

Time: 13:45 - 15.00
Room: Beam (60) 
Language: English
Chaired by prof. Paul Havinga

How can organizations share information securely without revealing any underlying sensitive data? That challenge is being addressed in this session. 

Sharing data between organizations can give deep insight into the hidden information between them. However, relevant data is often too sensitive to be shared with others, due to privacy concerns (e.g. GDPR), but also because of commercial sensitivity of the data. By securely combining and enriching available data, organizations can become more efficient and effective, and participating parties save a lot of money. It allows to better identify risks and threats, to improve healthcare, to better detect financial economic crimes, etc. Innovative ICT solutions that deal with this challenge include Secure Multi-Party Computation (MPC), Federated Learning and International Data Spaces. Specifically, MPC is a ‘toolbox’ of cryptographic techniques that allows several different parties to jointly analyze data, just as if they have a shared database. At the same time, the underlying sensitive data remains secure and only the output of the analysis will be revealed.
 
In this session TNO, jointly with partners, will present concrete applications of MPC and federated learning in healthcare and in financial crime detection, including a discussion on the broader opportunities and challenges of these techniques and related solutions.
 

  • Privacy friendly and data-driven HIV treatments. 
    Thomas Attema (TNO en CWI)
    The extreme variability in viral genotypes of HIV and the large number of antiretroviral drugs make the prescription of optimal treatments a complex task. To aid physicians in this task clinical decision support systems (CDSS) are used. However, mainly due to privacy and confidentiality constraints, the amount of data available to these systems is limited. Based on Secure Multi-Party Computation (MPC), TNO, CWI and UvA have developed a solution for CDSS that utilizes patient’s health records while preserving the privacy of these patients.
  • Enabling privacy-preserving analyses on federated healthcare data. 
    Dr. Gijs Geleijnse (IKNL) and dr. Daniël Worm (TNO)
    The growing complexity of cancer diagnosis and treatment requires data sets that are larger and richer than currently available in a single cancer registry. However, sharing patient data is difficult due to patient privacy and data protection needs. Privacy preserving distributed learning technologies like federated learning and MPC have the potential to overcome these limitations. IKNL and TNO are collaborating to develop solutions using these technologies to enable training of survival analysis models.
  • Detecting transaction fraud using Secure Multi-Party Computation.
    Mark Wiggerman (ABN AMRO) and Alex Sangers (TNO)
    Collaboration between financial institutions could improve the detection of transaction fraud. However, exchange of relevant data between these institutions is often not possible due to privacy constraints and data confidentiality. In a research project, TNO, ABN AMRO, ING and Rabobank developed a secure solution using MPC. This MPC solution would enable multiple banks to cooperatively build a fraud detection model that is based on the transaction data of the participating banks, without sharing any transaction data.
     
  • Q&A with all speakers.