Learning with Pseudo-Ensembles

TL;DR

This paper introduces the concept of pseudo-ensembles, a framework for creating and regularizing collections of models generated by noise, improving performance especially in semi-supervised learning scenarios.

Contribution

It formalizes pseudo-ensembles, proposes a novel regularizer for robustness, and demonstrates its effectiveness in both neural networks and sentiment analysis tasks.

Findings

Regularizer matches dropout performance in supervised learning

Extends naturally to semi-supervised learning with state-of-the-art results

Improves a recursive neural network's performance on sentiment analysis

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.