Zeroth-order hard-thresholding: gradient error vs. expansivity
W de Vazelhes, H Zhang, H Wu… - Advances in Neural …, 2022 - proceedings.neurips.cc
Abstract $\ell_0 $ constrained optimization is prevalent in machine learning, particularly for
high-dimensional problems, because it is a fundamental approach to achieve sparse …
high-dimensional problems, because it is a fundamental approach to achieve sparse …
[PDF][PDF] On Iterative Hard-Thresholding: Gradient Estimation and Non-Convex Projections Thesis Plan
W de Vazelhes - 2023 - wdevazelhes.github.io
Learning sparse models is an important topic in machine learning, in particular with the ever
increasing dimensionality of data. In this candidacy thesis, we extend the existing work on …
increasing dimensionality of data. In this candidacy thesis, we extend the existing work on …