English

Debiasing Stochastic Gradient Descent to handle missing values

Statistics Theory 2020-06-09 v2 Statistics Theory

Abstract

Stochastic gradient algorithm is a key ingredient of many machine learning methods, particularly appropriate for large-scale learning.However, a major caveat of large data is their incompleteness.We propose an averaged stochastic gradient algorithm handling missing values in linear models. This approach has the merit to be free from the need of any data distribution modeling and to account for heterogeneous missing proportion.In both streaming and finite-sample settings, we prove that this algorithm achieves convergence rate of O(1n)\mathcal{O}(\frac{1}{n}) at the iteration nn, the same as without missing values. We show the convergence behavior and the relevance of the algorithm not only on synthetic data but also on real data sets, including those collected from medical register.

Keywords

Cite

@article{arxiv.2002.09338,
  title  = {Debiasing Stochastic Gradient Descent to handle missing values},
  author = {Julie Josse and Aude Sportisse and Claire Boyer and Aymeric Dieuleveut},
  journal= {arXiv preprint arXiv:2002.09338},
  year   = {2020}
}
R2 v1 2026-06-23T13:49:30.714Z