Pushing the Frontiers of Federated Learning: From Security Applications to Mitigation of Poisoning Attack
- Ahmad-Reza Sadeghi (TU Darmstadt)
- Monday 28 November 2022
Niels Bohrweg 1
2333 CA Leiden
- Room 312
Federated Learning (FL) is a collaborative machine learning approach allowing several parties to jointly train a model without the need to share their private local datasets. FL is an enabling technology that can benefit distributed security-critical applications.
Recently, FL is shown to be susceptible to poisoning attacks, in which an adversary injects manipulated model updates into the federated model aggregation process to destroy or corrupt the resulting predictions, or implant hidden functionalities (aka backdoors). In this talk, we present our recent research work and experiences, also with industrial partners, concerning both the utilization of FL in large scale security applications as well as building FL systems resilient to poisoning attacks. Finally, we discuss the lessons learned and future research directions.
Ahmad-Reza Sadeghi is a professor of Computer Science and the head of the System Security Lab at Technical University of Darmstadt, Germany. He has been leading several Collaborative Research Labs with Intel since 2012, and with Huawei since 2019. For his influential research on Trusted and Trustworthy Computing he received the renowned German "Karl Heinz Beckurts" award that honors excellent scientific achievements with high impact on industrial innovations in Germany. In 2018, he received the ACM SIGSAC Outstanding Contributions Award for dedicated for pioneering contributions in content protection, mobile security and hardware-assisted security. In 2021, he was honored with Intel Academic Leadership Award at USENIX Security conference for his influential research on cybersecurity and in particular on hardware security. He is also the recipient of the prestigious European Research Council (ERC) Advanced Grant.