Teodora Reu

DPhil in Computer Science at University of Oxford

prof_pic_2026.jpg
prof_art.jpg
oil painting of a terrace by the sea
watercolour of colourful houses
oil still life with a gold pitcher
watercolour of mountains
surreal oil painting with a blue dog
oil painting of a flower bouquet
watercolour of a yellow wall with a red door
more paintings →

Hey! I’m Teo, a DPhil student at Oxford working with Professor Michael Bronstein on generative models (diffusion, flow matching, optimal transport) and geometric deep learning. I’ve published at NeurIPS 2024/2025 (Spotlight!) and UAI 2024, and recently interned at Wayve training large-scale video autoencoders for autonomous driving.

selected publications

  1. Gradient Variance Reveals Failure Modes in Flow-Based Generative Models
    Teodora Reu, Sixtine Dromigny, Michael Bronstein, and Francisco Vargas
    Advances in Neural Information Processing Systems (NeurIPS 2025, Spotlight), 2025
  2. Metric flow matching for smooth interpolations on the data manifold
    Kacper Kapuśniak, Peter Potaptchik, Teodora Reu, Leo Zhang, and 4 more authors
    Advances in Neural Information Processing Systems, 2024
  3. To smooth a cloud or to pin it down: Expressiveness guarantees and insights on score matching in denoising diffusion models
    Teodora Reu, Francisco Vargas, Anna Kerekes, and Michael Bronstein
    In Proceedings of the Fortieth Conference on Uncertainty in Artificial Intelligence (UAI 2024), 2024
  4. Topological message passing for higher-order and long-range interactions
    Lorenzo Giusti, Teodora Reu, Francesco Ceccarelli, Cristian Bodnar, and 1 more author
    In 2024 International Joint Conference on Neural Networks (IJCNN), 2024