Numerical optimization algorithms for convex and nonconvex problems.
On-line optimization
Recommender systems with tabular data
Software development: design principles, TDD and BDD, C++, C#, F#
Projects
Proximal point with exponential losses - a tiny Python library containing an fast and numerically accurate PyTorch and NumPy implementations of the proximal operator of exponential loss functions (Poisson Regression, Tilted Loss, etc)
Incremental proximal point - a library of algorithms for building incremental proximal-point methods for machine learning.
TREM OPF Solver - A MATLAB solver for a class of optimal power flow problems using the Tree Reduction and Expansion Method. See paper below.
AutoDiff - a .NET library for automatically computing derivatives of mathematical functions.
Publications
Dan Greenstein, Elazar Gershuni, Ilan Ben-Bassat, Yaroslav Fyodorov, Ran Moshe, Fiana Raiber, Alex Shtoff, Oren Somekh, Nadav Hallak A Stochastic Approach to the Subset Selection Problem via Mirror Descent. The Thirteenth International Conference on Learning Representations (ICLR 2025) (Paper)
Alex Shtoff, Elie Abboud, Rotem Stram, Oren Somekh Function Basis Encoding of Numerical Features in Factorization Machines. Transactions on Machine Learning Research (TMLR, 2024) (Paper, Code)
Alex Shtoff, Michael Viderman, Naama Haramaty Krasne, Oren Somekh, Ariel Raviv, Tularam Ban Low Rank Field-Weighted Factorization Machines for Low Latency Item Recommendation. 18th ACM Conference on Recommender Systems (RecSys 2024) (Paper), (Code)
Yohay Kaplan, Yair Koren, Alex Shtoff, Oren Somekh Conversion-Based Dynamic-Creative-Optimization in Native Advertising. 2022 IEEE International Conference on Big Data (BigData) (Preprint)
