I am a 4th year PhD student at the Cambridge Image Analysis group in the Department of Applied Mathematics and Theoretical Physics. I am also a member of the Cantab Capital Institute for the Mathematics of Information.
My research focuses on generative modeling, with a particular emphasis on diffusion models. Recently, I have become deeply interested in disentangled representation learning and the extraction of data manifold properties from pretrained generative models, including diffusion models and normalizing flows. Additionally, I have been exploring the fascinating area of Schrödinger Bridges.
If any of those topics sound interesting, reach out. I am always happy to chat!
Publications
- Your diffusion model secretly knows the dimension of the data manifold, Jan Stanczuk*, Georgios Batzolis*, Teo Deveney, Carola-Bibiane Schönlieb, arXiv preprint arXiv:2212.12611, 2022 [PDF] (Accepted at ICML 2024)
- CAFLOW: conditional autoregressive flows, Georgios Batzolis, Marcello Carioni, Christian Etmann, Soroosh Afyouni, Zoe Kourtzi, Carola-Bibiane Schönlieb, arXiv preprint arXiv:2106.02531, 2021 [PDF] (Accepted at Foundations of Data Science, AIMS Journal)
- How to distribute data across tasks for meta-learning?, Alexandru Cioba, Michael Bromberg, Qian Wang, Ritwik Niyogi, Georgios Batzolis, Jezabel Garcia, Da-shan Shiu, Alberto Bernacchia, Proceedings of the AAAI Conference on Artificial Intelligence, 2022 [PDF]
Preprints
- Variational Diffusion Auto-encoder: Latent Space Extraction from Pre-trained Diffusion Models, Georgios Batzolis*, Jan Stanczuk*, Carola-Bibiane Schönlieb, arXiv preprint arXiv:2304.12141, 2023 [PDF]
- Non-uniform diffusion models, Georgios Batzolis, Jan Stanczuk, Carola-Bibiane Schönlieb, arXiv preprint arXiv:2207.09786, 2022 [PDF]
- Conditional image generation with score-based diffusion models, Georgios Batzolis, Jan Stanczuk, Carola-Bibiane Schönlieb, Christian Etmann, arXiv preprint arXiv:2111.13606, 2021 [PDF]
* denotes equal contribution