Mikhail Belkin

Mikhail Belkin is a leading figure in the mathematics of machine learning and artificial intelligence. He received his Ph.D. in Mathematics from the University of Chicago in 2003, with a thesis titled 'Problems of Learning on Manifolds' under the supervision of Partha Niyogi. Prior to his doctoral studies, he earned an M.Sc. in Mathematics from the same institution in 1997 and a Hon.B.Sc. with High Distinction in Mathematics from the University of Toronto in 1995. Following his Ph.D., Belkin served as a Postdoctoral Researcher in the Department of Computer Science at the University of Chicago from December 2003 to August 2005. He then joined Ohio State University in 2005 as an Assistant Professor in the Department of Computer Science and Engineering, progressing to Associate Professor and Full Professor, while holding a courtesy appointment in the Department of Statistics until 2020. Since 2020, he has been Professor at the Halıcıoğlu Data Science Institute (HDSI) at the University of California, San Diego, where he holds the HDSI Endowed Chair in Artificial Intelligence and maintains an affiliate appointment in Computer Science and Engineering. He has also held visiting positions at institutions such as Simons Institute for the Theory of Computing, IST Austria, and SAMSI.

Belkin's research focuses on the mathematical, statistical, and computational foundations of deep learning, including spectral, manifold, and kernel methods, optimization, semi-supervised learning, clustering, and phenomena like overparameterization, interpolation, and double descent. He introduced foundational methods such as Laplacian Eigenmaps for dimensionality reduction and data representation (Neural Computation, 2003), Manifold Regularization: A geometric framework for learning from labeled and unlabeled examples (Journal of Machine Learning Research, 2006), and Consistency of spectral clustering (Conference on Learning Theory, 2008). Influential recent works include Reconciling modern machine-learning practice and the classical bias–variance trade-off (PNAS, 2019), Fit without fear: remarkable mathematical phenomena of deep learning through the prism of interpolation (Acta Numerica, 2021), Mechanism for feature learning in neural networks and backpropagation-free machine learning models (Science, 2024), and Linear recursive feature machines provably recover low-rank matrices (PNAS, 2025). His publications have amassed tens of thousands of citations, shaping modern understanding of machine learning theory. Belkin has received the NSF CAREER Award (2007), Lumley Research Award (2011), and was elected an ACM Fellow (2023) for contributions to modern machine learning theory and algorithms. He currently serves as Editor-in-Chief of SIAM Journal on Mathematics of Data Science (since 2024), previously as Action Editor for Journal of Machine Learning Research and Associate Editor for IEEE TPAMI, and Program Chair for COLT 2021.