A Theoretically Grounded Application of Dropout in Recurrent Neural Networks

NeurIPS 2016  ·  Yarin Gal, Zoubin Ghahramani ·

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.

PDF Abstract NeurIPS 2016 PDF NeurIPS 2016 Abstract

Datasets


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Language Modelling Penn Treebank (Word Level) Gal & Ghahramani (2016) - Variational LSTM (large) Validation perplexity 77.9 # 27
Test perplexity 75.2 # 34
Language Modelling Penn Treebank (Word Level) Gal & Ghahramani (2016) - Variational LSTM (medium) Validation perplexity 81.9 # 28
Test perplexity 79.7 # 37

Methods