I am a professor at University of Neuchatel and TU Delft. I am the director of Distributed Learning Systems Lab. I returned to academia, after a decade of industry experience at the IBM Research Zurich Lab. My research interests lie in the distinct areas of deep machine learning, big data systems, and privacy enhancing technology. I am also the co-founder of BlueGen.ai. My research is supported by the Swiss National Science Foundation, Dutch National Science Foundation the European Union, IBM Research, ABB, and TU Delft.

I have multiple open PhD positions. If you are interested in them, please drop me an email at lydiaychen@ieee.org.

  • Privacy-preserving learning systems by synthetic data: how to maximize the knowledge of while maintaining data privacy ? I am exploring the deep generative models to synthesize data.
  • Robust learning systems: how to make learning algorithms robust against adversaries that maliciously manipulate data input? I am designing practical strategies and theories against attackers, e.g., free riders, and thief.
  • Federated machine learning systems: how to decentralizedly learn deep learning models on heterogeneous clients who contentiously encounter new learning tasks and data domain shift? I am working on domain adaptation, and continual learning in FL.


    • News

  • [07/23] Lydia will serve NWO Open Competition assessment committee
  • [06/23] Our paper, “Exploring and Exploiting Data-Free Model Stealing" , is accepted in ECML 23
  • [04/23] Our paper, “LeadFL: Client Self-Defense against Model Poisoning in Federated Learning" , is accepted in ICML 23
  • [03/23] Our paper, “Fabricated Flips: Poisoning Federated Learning without Data" , is accepted in DSN 23
  • [02/23] Our paper, “Maverick Matters: Client Contribution and Selection in Federated Learning" , is accepted in PAKDD 23
  • [01/23] Our paper, “Defending Against Free-Riders Attacks in Distributed Generative Adversarial Networks" , is accepted in Financial Crypto 23
  • [12/22] Our paper, “Robust Learning via Golden Symmetric Loss of (un)Trusted Labels " , is accepted in SIAM SDM 23
  • [10/22] Our paper, “FedKNow:Federated Continual Learning with Signature Task Knowledge Integration at Edge" , is accepted in IEEE ICDE 23

    • Old News.....