The heavy-tail phenomenon in SGD. M Gurbuzbalaban, U Simsekli, L Zhu. International Conference on Machine Learning, 3964-3975, 2021. 98, 2021 ; Stochastic Quasi- ...
My research field is best described as Mathematical Machine Learning*. I am/have been an area chair at NeurIPS, ICML, COLT, AISTATS, ICASSP, and WASPAA, and I ...
A tail-index analysis of stochastic gradient noise in deep neural networks. U Simsekli, L Sagun, M Gurbuzbalaban. International Conference on Machine Learning, ...
I find it very weird that in math it's super common to cite a mathematician's nationality along with their name, like "Chinese" or "Hungarian" or whatsoever. It's quite disappointing to be so nationalistic in a scientific field which is supposed to transcend time/space/nations
Twitter • 10 days ago
Last week I gave a lecture series on heavy tails in ML as part of a Newton Institute event www.newton.ac.uk/event/tm…. To my surprise one of my personal heros stat.cornell.edu/people/f… was in the audience. I learned most of heavy tails from his books, like many other people.
Twitter • 10 days ago
My neighbors are complaining again.
Twitter • 29 days ago
As opposed to the past research which has mainly focused on developing generic, instrument-independent pitch tracking methods, our models are instrument- ...
The codebase for the paper "A Tail-Index Analysis of Stochastic Gradient Noise in Deep Neural Networks". Python 21 10.
I am a Research Faculty (Chargé de Recherche) at INRIA – SIERRA team and École Normale Supérieure, Computer Science Department. I received my PhD from Boğaziçi ...
Publications · U. Şimşekli, “Tensor Fusion: Learning In Heterogeneous and Distributed Data”, PhD Thesis, Boğaziçi University, Istanbul, Turkey, 2015 · U.
Promoting openness in scientific communication and the peer-review process.
Umut Şimşekli is a Research Faculty at INRIA - SIERRA team and École Normale Supérieure, Computer Science Department. He is the 2021 winner of an ERC ...