Peter Potaptchik
University of Oxford
peter.potaptchik@stats.ox.ac.uk
I’m a second-year DPhil student in the Department of Statistics at the University of Oxford, under the supervision of George Deligiannidis, Saifuddin Syed and Yee Whye Teh. My studies are funded by the StatML CDT.
My research centers on generative modelling and sampling, with a focus on bridging various fields. I’m particularly interested in using techniques from differential geometry, optimal transport, probability, and statistical learning theory to deepen our understanding of the strengths and limitations of algorithms in these areas. I also have a broader interest in machine learning theory.
Before joining Oxford, I worked with Chris J. Maddison and Daniel Roy at the University of Toronto, where I earned my BSc in Computer Science and Statistics.
Feel free to reach out if you’d like to collaborate or just chat!
News
| Oct 17, 2024 | Our preprint Linear Convergence of Diffusion Models Under the Manifold Hypothesis is out. In the context of ImageNet, our work suggests using ~100 steps (intrinsic dim) for effective sampling, compared to the existing state-of-the-art bounds, which recommend ~150k steps (ambient dim). |
|---|---|
| Sep 25, 2024 | Our paper Metric Flow Matching for Smooth Interpolations on the Data Manifold was accepted at NeurIPS 2024. |
| Dec 26, 2023 | Our preprint de Finetti’s theorem and the existence of regular conditional distributions and strong laws on exchangeable algebras is out on arxiv. |
| Nov 27, 2023 | My talk at BAYSM (j-ISBA’s meeting) won the short talk award. |
| Sep 24, 2023 | I started my DPhil at the University of Oxford! |