Samet Oymak
 |
Samet Oymak Assistant Professor Electrical and Computer Engineering University of California, Riverside Office: Winston Chung Hall 322 Email: lastname (at) ece (dot) ucr (dot) edu |
I am an Assistant Professor of ECE at UC Riverside. I received my PhD degree in Electrical Engineering from Caltech in 2015.
Research: My research lies at the confluence of machine learning, optimization, and statistics. I am broadly interested in finding principled solutions to contemporary machine learning and data science problems by using tools from mathematical optimization and statistics. I also explore the fundamental algorithmic tradeoffs between speed, accuracy, scalability, and the data-efficiency.
Active research topics: (non)convex optimization, reinforcement learning, deep networks, and high-dimensional problems.
I am looking for enthusiastic Master's and undergraduate students who are interested in applied machine learning and data science projects.
Basic Python knowledge is necessary. Please reach out with a CV and transcript!
News
6/2019, new preprint on the generalization properties of neural nets.
6/2019, two workshop papers will be presented at ICML.
5/2019, new preprint with Yahya: “Quickly Finding the Best Linear Model in High Dimensions”.
4/2019, one paper accepted to ICML 2019.
4/2019, one paper accepted to COLT 2019.
4/2019, one France trip, two birds: papers accepted to SampTA 2019 and ISIT 2019.
4/2019, gave a talk at UCR Econometrics Seminar.
3/2019, new preprint with Mingchen and Mahdi: “Gradient Descent with Early Stopping is Provably Robust to Label Noise for Overparameterized Neural Networks”.
3/2019, will give a talk in Data Science Seminar at UCR.
2/2019, new preprint with Mahdi: “Towards moderate overparameterization: global convergence guarantees for training shallow neural networks”.
2/2019, will give a talk at ITA on recent results.
1/2019, two papers accepted to ICASSP 2019 and ACC 2019.
12/2018, new preprint with Mahdi: “Overparameterized Nonlinear Learning: Gradient Descent Takes the Shortest Path?”.
09/2018, new preprint “Stochastic gradient descent learns state equations with nonlinear activations”.
09/2018, gave an invited talk at University of Michigan.
07/2018, presented a paper in ICML 2018.
06/2018, gave an invited talk at Bilkent University.
06/2018, new preprint with Necmiye, “Non-asymptotic Identification of LTI Systems from a Single Trajectory”.
05/2018, new preprint with Mahdi: “End-to-end Learning of a Convolutional Neural Network via Deep Tensor Decomposition”.
05/2018, new preprint with Amir, Arindam, Kevin: “High Dimensional Data Enrichment: Interpretable, Fast, and Data-Efficient”.
05/2018, gave an invited talk at UCR Data Science Seminar.
02/2018, started UC Riverside!
02/2018, new preprint “Learning Compact Neural Networks with Regularization”.