Jiyue Huang

Jiyue Huang

Ph.D student now exploring Privacy and Security of Heterogeneous Distributed Learning Systems.

Teaching

Guest lecturer of 62122 distributed deep learning systems

Master course, Univerisity of Neuchatel, 2024

Guest lecturer for FS2024: 62122 Distributed Deep Learning Systems at the University of Neuchatel on the topic of advanced attacks in distributed learning systems. The lecture includes concepts and preliminaries in attacks and defenses of distributed learning systems. Examples of advanced privacy leakage as well as security risks are discussed. A research-oriented group task of improving gradient inversion attacks is designed.

Supervisor for master thesis

Master thesis, Department of EEMCS, TU Delft, 2024

Ongoing supervising master student Caspar for thesis for 6 months, collaborated with TNO, on the topic of gradient inversion attacks on Time-series energy data.

Supervisor of CSE3000

Bachelor course, Department of EEMCS, TU Delft, 2023

Supervised 4 bachelor students (Todor Mladenović, Quinten Van Opstal, Jan van der Meulen, and Lazar Nenovski) for 3 months, on the course of Malicious Parties and Defenses in Multi-Server Federated Learning with four good thesis projects.

Guest Lecturer and TA for CS4290

Master course, Department of EEMCS, TU Delft, 2022

Guest Lecturer for CS4290 Seminar on Distributed Machine Learning on the topic of malicious behaviors in Federated Learning. Teaching Assistant for CS4290 Seminar for two years, including choosing research paper for students to review, and grading accordingly.

Co-supervisor for bachelor thesis

Bachelor thesis, Department of EEMCS, TU Delft, 2022

Co-supervising 5 bachelor students (Kanish Dwivedi, Joost Jansen, Pietro Vigilanza Lorenzo, Steffano Psathas and Floris van Veen) for thesis for 3 months, including task explanation, research question formalization, QA during solutions, thesis writing and defence preparation.

Co-supervisor for master thesis

Master thesis, Department of EEMCS, TU Delft, 2021

Co-supervising master student Jin Xu for thesis for 6 months, including topic choosing, research question formalization, possible solution and paper writing. This gradient inversion attack of Federated Learning is published on SRDS 2022.