IMPORTANT: To comply with the coronavirus (COVID-19) safety measures, this event will take place online. The agenda and schedule will be adjusted to fit the online environment and will be available soon.
On Wednesday April 15, we organize a Techruption ideation session concerning secure, distributed learning as a start for a new use case. Secure, distributed learning was mentioned several times during the MPCL stagegates + brainstorm session on Dec 4th, 2019 and as such we warmly invite all interested partners (also the newer members!) to join a dedicated ideation session.
If you are interested in joining this ideation session, then please register via this page, before April 10th! It is assumed that participants are familiar with the concept of secure, distributed learning (click here to go to the preliminary event for an introductory session).
Data is the new oil. It fuels your organization with new insights (opportunities, efficiency etc.) after careful purification and analyses (e.g. artificial intelligence). There is even more to be gained through data collaborations with other organizations (e.g. complementary data), but there are many reasons for you to be unwilling or unable to share data. By secure distributed learning, we think of the application of several (fairly) novel techniques such as federated learning (FL) and secure multi-party computation (MPC) to collaboratively train or exploit e.g. machine learning models on distributed datasets without disclosing your raw data or other intermediate results.
During this ideation, we intent to identify interesting opportunities/questions that concern secure, distributed learning. In contrast to the ‘regular’ use cases, we intend to start a more technically-oriented use case; that is, there will be less focus on one agreed (business) application and more focus on deepening knowledge and experience with these enabling technologies. Typical questions that you may think of are:
- Which AI techniques can we already apply to distributed, sensitive data?
- Which tools or platforms currently support such secure analyses and how do they compare?
- What are the limitations and advantages of different approaches (e.g. FL versus MPC).
- What are relevant applications of secure, distributed learning? What are the needs and requirements of involved organizations?
Potential deliverable of use case: white paper that presents outcome of (some of the above) use case questions.
Participants are invited to prepare the ideation by generating ideas a priori so that we can discuss the ideas effectively. We conclude by evaluating the participants interest for a follow-up and defining the next steps.
Everyone is welcome to attend; however, a somewhat technical background and/or knowledge about the current status of your organization’s position regarding AI would help us to get the most out of this session. It is assumed that participants are familiar with the concept of secure, distributed learning (click here to go to the preliminary event for an introductory session).