Learning with Pseudo-Ensembles
Abstract
We formalize the notion of a pseudo-ensemble, a (possibly infinite) collection of child models spawned from a parent model by perturbing it according to some noise process. E.g., dropout (Hinton et. al, 2012) in a deep neural network trains a pseudo-ensemble of child subnetworks generated by randomly masking nodes in the parent network. We present a novel regularizer based on making the behavior of a pseudo-ensemble robust with respect to the noise process generating it. In the fully-supervised setting, our regularizer matches the performance of dropout. But, unlike dropout, our regularizer naturally extends to the semi-supervised setting, where it produces state-of-the-art results. We provide a case study in which we transform the Recursive Neural Tensor Network of (Socher et. al, 2013) into a pseudo-ensemble, which significantly improves its performance on a real-world sentiment analysis benchmark.
Cite
@article{arxiv.1412.4864,
title = {Learning with Pseudo-Ensembles},
author = {Philip Bachman and Ouais Alsharif and Doina Precup},
journal= {arXiv preprint arXiv:1412.4864},
year = {2014}
}
Comments
To appear in Advances in Neural Information Processing Systems 27 (NIPS 2014), Advances in Neural Information Processing Systems 27, Dec. 2014