avatar

Sofiane ENNADIR

AI Researcher
Microsoft (ABK - King AI Labs)
ennadir [at] kth [dot] se


About Me

I am an AI/ML Researcher at Microsoft (King AI Labs), where my work primarily focuses on Temporal Graph Neural Networks and their application across various industrial use cases. Recently, my interests have expanded to include the inner workings of Transformer-based architectures and Large Language Models (LLMs), aiming to better understand their underlying mechanisms.

I am currently finalizing my PhD in the School of Electrical Engineering and Computer Science at the KTH Royal Institute of Technology in Sweden. My research is conducted under the supervision of Professor Michalis Vazirgiannis and Professor Henrik Boström, with funding from the Wallenberg AI, Autonomous Systems and Software Program (WASP). My doctoral work explores the robustness of Graph Neural Networks (GNNs), with a particular emphasis on adversarial attacks. I aim to both analyze and develop adversarial strategies, while designing efficient and practical defense mechanisms grounded in theoretical insights.

I had the opportunity to spend a summer at the Flatiron Institute as part of the Foundation Models for Science initiative, where I collaborated with Leopoldo Sarra and Siavash Golkar. Our work centered on extending the Joint Embedding approach to enable foundational modeling for time series data.

Before joining KTH, I earned an MSc in "Applied Mathematics – Data Sciences" from École Polytechnique in Paris, France. I also hold an Engineering Master's degree from EMINES, School of Industrial Management at Mohammed VI Polytechnic University (UM6P) in Morocco.

My academic CV can be accessed here.

News

Publications

* is used to denote equal contribution.

  1. Expressivity of Representation Learning on Continuous-Time Dynamic Graphs: An Information-Flow Centric Review
    S. Ennadir, G. Gandler, F. Cornell, L. Cao, O. Smirnov & Al.
    Transactions on Machine Learning Research (TMLR).
    PDF Code
  2. If You Want to Be Robust, Be Wary of Initialization
    S. Ennadir, J. Lutzeyer, M. Vazirgiannis & E. Bergou.
    The Thirty-Eighth Annual Conference on Neural Information Processing Systems (Neurips), 2024.
    PDF Code BibTex
  3. Joint Embedding go Temporal
    S. Ennadir, S. Golkar, L. Sarra.
    Time Series in the Age of Large Models Workshop, Neurips 2024.
    PDF Code
  4. Bounding the Expected Robustness of Graph Neural Networks Subject to Node Feature Attacks
    Y. Abbahadou*, S. Ennadir*, J. Lutzeyer, M. Vazirgiannis & H. Boström.
    The Twelfth International Conference on Learning Representations (ICLR), 2024.
    PDF Code BibTex

  5. A Simple and Yet Fairly Effective Defense for Graph Neural Networks
    S. Ennadir , Y. Abbahadou, J. Lutzeyer, M. Vazirgiannis & H. Boström.
    Thirty-Eighth AAAI Conference on Artificial Intelligence (AAAI), 2024.
    PDF Code BibTex
    [Previous version] presented at the ICML 2nd AdvML workshop 2023.

  6. Interpretable Graph Neural Networks for Tabular Data
    A. Alkhatib, S. Ennadir, H. Boström & M. Vazirgiannis.
    The 27th European Conference on Artificial Intelligence (ECAI), 2024.
    PDF Code
    [Previous version] presented at the DMLR workshop in ICLR 2024.

  7. UnboundAttack : Generating Unbounded Adversarial Attacks to Graph Neural Networks
    S. Ennadir, A. Alkhatib, G. Nikolentzos, M. Vazirgiannis & H. Boström.
    12th International Conference on Complex Networks (CNA), 2023.
    PDF Slides
  8. Structure-Aware Antibiotic Resistance Classification using Graph Neural Networks
    A. Qabel, S. Ennadir, G. Nikolentzos, J. Lutzeyer, M. Chatzianastasis, H. Boström & M. Vazirgiannis.
    AI4Science Workshop at Neurips, 2022.
    PDF Code BibTex
    [Extended version] currently under review.



I would like to thank Johannes Lutzeyer for allowing me to copy the format of his website. This website is powered by Jekyll and Minimal Light theme.