Learning Mixtures of Linear Classifiers
Machine Learning
2014-08-01 v4 Machine Learning
Abstract
We consider a discriminative learning (regression) problem, whereby the regression function is a convex combination of k linear classifiers. Existing approaches are based on the EM algorithm, or similar techniques, without provable guarantees. We develop a simple method based on spectral techniques and a `mirroring' trick, that discovers the subspace spanned by the classifiers' parameter vectors. Under a probabilistic assumption on the feature vector distribution, we prove that this approach has nearly optimal statistical efficiency.
Cite
@article{arxiv.1311.2547,
title = {Learning Mixtures of Linear Classifiers},
author = {Yuekai Sun and Stratis Ioannidis and Andrea Montanari},
journal= {arXiv preprint arXiv:1311.2547},
year = {2014}
}