Belkin Lab @ UCSD

Research

We are a research group working on understanding modern machine learning, and especially deep learning. We take a scientific approach, using tools across the spectrum from theory to experiment. For example, we want to characterize the behavior of deep learning systems, understand which design choices are necessary or sufficient for these behaviors, and identify universal structures underlying machine learning methods (deep or otherwise).

We are particularly interested in the areas of generalization, interpolation, optimization, overparameterization, and relations between these objects. In the past, our group has published influential results in Double Descent, Neural Tangent Kernels, overparameterization, and frameworks for generalization.

If Deep Learning is the solution, what is the problem?

People

Faculty:

Mikhail Belkin

Current PhD students:

Amirhesam Abedsoltan
Daniel Beaglehole
Fivos Kalogiannis
Akash Kumar (co-advised by Sanjoy Dasgupta)
Neil Mallinar
Parsa Mitraheri

Current Postdocs:

Qingsong Wang

Former Group Members:

Like Hui (currently a Researcher at Yahoo)
Chaoyue Liu (currently an Assistant Professor at Purdue University)
Siyuan Ma (currently a Software Engineer at Google)
Preetum Nakkiran (currently a Research Scientist at Apple)
Parthe Pandit (currently an Assistant Professor at IIT Bombay)
Libin Zhu (currently a Postdoc at University of Washington)

Collaborators:

Enric Boix-Adsera (currently an Assistant Professor at University of Pennsylvania)
Nikhil Ghosh (currently a Research Fellow at the Flatiron Institute)
Adit Radhakrishnan (currently an Assistant Professor at MIT Math)
James B. Simon (currently a Research Fellow at Imbue and Research Scientist at the Redwood Center)

Lab social, June 2025, HDSI Building at UCSD

Lab meeting, September 2021