Robustly Learning a Single Neuron via Sharpness
Machine Learning
2023-06-14 v1 Data Structures and Algorithms
Optimization and Control
Statistics Theory
Machine Learning
Statistics Theory
Abstract
We study the problem of learning a single neuron with respect to the -loss in the presence of adversarial label noise. We give an efficient algorithm that, for a broad family of activations including ReLUs, approximates the optimal -error within a constant factor. Our algorithm applies under much milder distributional assumptions compared to prior work. The key ingredient enabling our results is a novel connection to local error bounds from optimization theory.
Cite
@article{arxiv.2306.07892,
title = {Robustly Learning a Single Neuron via Sharpness},
author = {Puqian Wang and Nikos Zarifis and Ilias Diakonikolas and Jelena Diakonikolas},
journal= {arXiv preprint arXiv:2306.07892},
year = {2023}
}