A Theoretically Grounded Application of Dropout in Recurrent Neural Networks

TL;DR

This paper provides a Bayesian interpretation of dropout, extending it to recurrent neural networks, and demonstrates improved performance in language modeling and sentiment analysis tasks.

Contribution

It introduces a variational inference-based dropout method for RNNs, improving over existing techniques and achieving state-of-the-art results.

Findings

Outperforms existing dropout methods in RNNs

Achieves state-of-the-art perplexity on Penn Treebank

Enhances the theoretical understanding of dropout in RNNs

Abstract

Recurrent neural networks (RNNs) stand at the forefront of many recent developments in deep learning. Yet a major difficulty with these models is their tendency to overfit, with dropout shown to fail when applied to recurrent layers. Recent results at the intersection of Bayesian modelling and deep learning offer a Bayesian interpretation of common deep learning techniques such as dropout. This grounding of dropout in approximate Bayesian inference suggests an extension of the theoretical results, offering insights into the use of dropout with RNN models. We apply this new variational inference based dropout technique in LSTM and GRU models, assessing it on language modelling and sentiment analysis tasks. The new approach outperforms existing techniques, and to the best of our knowledge improves on the single model state-of-the-art in language modelling with the Penn Treebank (73.4 test perplexity). This extends our arsenal of variational tools in deep learning.