Adversarial Dropout for Recurrent Neural Networks

Authors

  • Sungrae Park NAVER Corporation
  • Kyungwoo Song KAIST
  • Mingi Ji KAIST
  • Wonsung Lee KAIST
  • Il-Chul Moon KAIST

DOI:

https://doi.org/10.1609/aaai.v33i01.33014699

Abstract

Successful application processing sequential data, such as text and speech, requires an improved generalization performance of recurrent neural networks (RNNs). Dropout techniques for RNNs were introduced to respond to these demands, but we conjecture that the dropout on RNNs could have been improved by adopting the adversarial concept. This paper investigates ways to improve the dropout for RNNs by utilizing intentionally generated dropout masks. Specifically, the guided dropout used in this research is called as adversarial dropout, which adversarially disconnects neurons that are dominantly used to predict correct targets over time. Our analysis showed that our regularizer, which consists of a gap between the original and the reconfigured RNNs, was the upper bound of the gap between the training and the inference phases of the random dropout. We demonstrated that minimizing our regularizer improved the effectiveness of the dropout for RNNs on sequential MNIST tasks, semi-supervised text classification tasks, and language modeling tasks.

Downloads

Published

2019-07-17

How to Cite

Park, S., Song, K., Ji, M., Lee, W., & Moon, I.-C. (2019). Adversarial Dropout for Recurrent Neural Networks. Proceedings of the AAAI Conference on Artificial Intelligence, 33(01), 4699-4706. https://doi.org/10.1609/aaai.v33i01.33014699

Issue

Section

AAAI Technical Track: Machine Learning