Paper accepted at KBS
The paper proposes an anonymization method that can publish finite microdata streams with high protection as well as high utility. Download here.
My research attempts to leverage mature concepts and techniques amply utilized in information theory, convex optimization and stochastic estimation, to tackle the increasingly controversial problem of privacy in modern information systems from a perspective that is mathematically systematic and adheres to the principles of engineering optimization.
My current research interests encompass (i) the anonymization of static and dynamic data with syntactic and differential privacy guarantees; (ii) the generation of differentially-private synthetic data; and (iii) the design of mechanisms that allow machine learning algorithms to learn accurate models while protecting the privacy of the individuals on whom the data are trained.
The goal is to develop privacy-preserving, artificially-intelligence-based analytics for smart cities. We aim to enable stakeholders to operate on protected data while ensuring those data are useful for analytics.
In the context of physical training and rehabilitation, we aim to investigate to which extent our physiologycal behavior, and more specifically our muscle activation patterns, are unique and therefore can be used for reidentification.
The paper proposes an anonymization method that can publish finite microdata streams with high protection as well as high utility. Download here.
All project members have met at SAP France. Really interesting discussions on how to protect location data in smart cities. Stay tuned!
Our work presented at the IEEE Conference on Dependable and Secure Computing (DSC'21) receives the best-paper award. Download here.
Please send us your query and we will get back to you as soon as possible