Megha Khosla

Assistant Professor, Delft University of Technology (TU Delft)

prof_pic.jpg

Van Mourik Broekmanweg 6

2628 XE Delft

The Netherlands

I am an Assistant Professor in the Multimedia Computing Group in the Intelligent Systems Department at TU Delft. My research focuses on trustworthy machine learning, grounded in graph-structured data.

I lead a research line on the interplay between the transparency and privacy needs of AI and machine learning models. These needs can appear contradictory: understanding a model calls for disclosure, while protecting sensitive information calls for restraint. I see this apparent contradiction as an invitation to understand more deeply what transparency requires, what privacy protects, and where the two can reinforce each other. My recent Vidi project supports my investigation of focused questions about this interplay.

Prior to my being a faculty member at TUD, I was a senior researcher at the L3S Research Centre, and the Leibniz University Hannover. I have also managed several collaborative projects both in Academia and in the industry where I spent a three-year stint after my PhD. I completed my PhD from the Max Planck Institute for Informatics (MPII), in Algorithms and the Complexity group, Saarbruecken, Germany.

news

May 2026 Congratulations to Tianqi Zhao on defending her PhD thesis in May 2026. Check out her thesis, Characterizing learning difficulty in graph-structured data: an empirical study of models and data.
October 2025 Thrilled to share that I received an NWO Vidi grant for PriXAI: A Paradigm Shift in Explainable AI for Privacy and Decision Support.

selected publications

  1. Impact of Graph Structure on Membership-Inference Risk for Graph Neural Networks
    Khosla, Megha
    Proceedings on Privacy Enhancing Technologies 2026
    HTML
  2. Population-Scale Network Embeddings Expose Educational Divides in Network Structure Related to Right-Wing Populist Voting
    Lüken, Malte, Garcia-Bernardo, Javier, Deb, Sreeparna, Hafner, Flavio, and Khosla, Megha
    Scientific Reports 2026
    arXiv HTML
  3. Disentangled and Self-Explainable Node Representation Learning
    Piaggesi, Simone, Panisson, André, and Khosla, Megha
    Transactions on Machine Learning Research 2025
  4. Multi-label Node Classification On Graph-Structured Data
    Zhao, Tianqi, Dong, Ngan Thi, Hanjalic, Alan, and Khosla, Megha
    In Transactions on Machine Learning Research 2023
  5. Private Graph Extraction via Feature Explanations
    Olatunji, Iyiola E, Rathee, Mandeep, Funke, Thorben, and Khosla, Megha
    In Proceedings of Privacy Enhancing Technologies Symposium (PETS 2023) 2023
  6. A message passing framework with multiple data integration for miRNA-disease association prediction
    Dong, Thi Ngan, Johanna, Schrader, Mucke, Stefanie, and Khosla, Megha
    Nature Scientific Reports 2022
  7. ZORRO: Valid, Sparse, and Stable Explanations in Graph Neural Networks
    Funke, Thorben, Khosla, Megha, Rathee, Mandeep, and Anand, Avishek
    IEEE Transactions on Knowledge and Data Engineering 2022