Privacy in data and in machine learning models

When sharing personal data publicly, it is promised that identity of the people whose data is published will remain hidden. Anonymization is necessary but not enough, such that there are many examples of re-identification on anonymized databases. The issue is not limited to the data, but also to models as well when shared publicly and trained on personal data with potentially sensitive contents such as credit card numbers or industrial strategies. Such discussions are the main topics in the context of privacy. Speaker(s): Mehdi Amian, Virtual: https://events.vtools.ieee.org/m/287722