Jacob Bamberger

Bonjour / Hej / Hi!

I am Jacob, a DPhil candidate in Computer Science at University of Oxford, supervised by Professor Michael Bronstein and Professor Xiaowen Dong. My research explores how tools from geometry, topology, and algebra – particularly differential geometry – can be used to tackle problems in modern deep learning, such as graph neural networks and generative models.

My current research focuses on neural diffusion geometry: learning the intrinsic geometry of data with scalable neural models. Drawing on diffusion geometry, we develop training objectives that enable neural networks to estimate geometric quantities directly from data, as in our work on Riemannian Metric Matching. I am particularly interested in using this learned geometry to guide learning and generation when data are scarce, as is often the case in scientific applications.

Before Oxford, I completed an MSc in Computer Science at EPFL. In a previous life, I was an aspiring mathematician: I earned a BSc and an MSc in Mathematics from McGill, where I focused on geometric group theory under the supervision of Professor Daniel Wise.

Along the way, I have also interned in various tech companies and startups, including Microsoft, Giotto.ai, and Oracle Labs.