Adversarial Dropout for Recurrent Neural Networks
This work addresses the need for better regularization in RNNs for sequential data processing, but it appears incremental as it builds on existing dropout techniques by incorporating adversarial concepts.
The paper tackled the problem of improving generalization in recurrent neural networks (RNNs) by proposing adversarial dropout, which intentionally disconnects dominant neurons to enhance dropout effectiveness. The result showed improved performance on sequential MNIST, semi-supervised text classification, and language modeling tasks, though no concrete numbers were provided.
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.