Zepeng Zhang

prof_pic.jpg

I am a PhD student at École Polytechnique Fédérale de Lausanne (EPFL) 🇨🇭 working under the supervision of Prof. Olga Fink. During my PhD, I have also worked with Prof. Jhony H. Giraldo and Aref Einizade at Télécom Paris, Institut Polytechnique de Paris 🇫🇷 as a visiting PhD student.

My research focuses on generative modeling, graph machine learning, and graph signal processing. I develop methods that incorporate relational structure and optimization principles to improve inference, generalization, robustness, and computational efficiency. Currently, I am particularly interested in understanding the behavior and developing more efficient/effective flow matching methods for spatiotemporal systems.

Previously, I received my master’s degree at ShanghaiTech University 🇨🇳 and B.Eng. degree from Wuhan University (WHU) 🇨🇳. I also had the opportunity to visit Peking University (PKU) 🇨🇳 and City University of Hong Kong (CityU) 🇭🇰 as a research intern.

I am always open to potential collaborations. Please feel free to drop me an email if you’re interested.

News

May 26, 2026 Our paper “MM-Net: Recurrent Unfolding with Adaptive Majorization for Weighted Sum-Rate Beamforming” has been accepted by IEEE Signal Processing Letters.
May 04, 2026 Our paper “Spatiotemporal Imputation with Graph-Informed Flow Matching” has been accepted by ICML 2026.
Oct 19, 2025 Our paper “RINS-T: Robust Implicit Neural Solvers for Time Series Linear Inverse Problems” has been accepted by IEEE Transactions on Instrumentation & Measurement.
Sep 18, 2025 Our paper “GeoLink: Empowering Remote Sensing Foundation Model with OpenStreetMap Data” has been accepted by NeurIPS 2025.
Jul 24, 2025 Our paper “Algorithm-Informed Graph Neural Networks for Leakage Detection and Localization in Water Distribution Networks” has been accepted by Reliability Engineering & System Safety.
May 20, 2025 Our paper “Weighted Sum-Rate Maximization using MM: Convergence Rate and Deep Adaptation” has been accepted by EUSIPCO 2025.
Apr 13, 2025 I have joined Prof. Jhony H. Giraldo’s group at Télécom Paris, Institut Polytechnique de Paris as a visiting student.
Dec 09, 2024 Our work on “Domain Adaptive Unfolded Graph Neural Networks” has been accepted by AAAI 2025.
Sep 12, 2024 Our work on “Graph Neural Networks with Adaptive Structures” has been accepted by IEEE Journal of Selected Topics in Signal Processing.

Selected Publications

  1. ICML
    Spatiotemporal Imputation with Graph-Informed Flow Matching
    In International Conference on Machine Learning, 2026
  2. arXiv
    Graph-Spectral Flow Matching for Multivariate Time Series Anomaly Detection
    Zepeng Zhang, Jhony H. Giraldo, Wenbin Wang, and Olga Fink
    arXiv preprint arXiv:2609.36765, 2026
  3. AAAI
    Domain Adaptive Unfolded Graph Neural Networks
    Zepeng Zhang and Olga Fink
    In Proceedings of the AAAI Conference on Artificial Intelligence, 2025
  4. JSTSP
    Graph Neural Networks With Adaptive Structures
    Zepeng Zhang, Songtao Lu, Zengfeng Huang, and Ziping Zhao
    IEEE Journal of Selected Topics in Signal Processing, 2024