Robust Multilingual Part-of-Speech Tagging via Adversarial Training

Abstract

Adversarial training (AT) is a powerful regularization method for neural networks, aiming to achieve robustness to input perturbations. Yet, the specific effects of the robustness obtained from AT are still unclear in the context of natural language processing. In this paper, we propose and analyze a neural POS tagging model that exploits AT. In our experiments on the Penn Treebank WSJ corpus and the Universal Dependencies (UD) dataset (27 languages), we find that AT not only improves the overall tagging accuracy, but also 1) prevents over-fitting well in low resource languages and 2) boosts tagging accuracy for rare / unseen words. We also demonstrate that 3) the improved tagging performance by AT contributes to the downstream task of dependency parsing, and that 4) AT helps the model to learn cleaner word representations. 5) The proposed AT model is generally effective in different sequence labeling tasks. These positive results motivate further use of AT for natural language tasks.

NAACL 2018

References in corpus (6)

• Natural Language Processing (almost) from Scratch
• On the difficulty of training Recurrent Neural Networks
• Delving into Transferable Adversarial Examples and Black-box Attacks
• Adversarial Examples for Evaluating Reading Comprehension Systems
• Data Noising as Smoothing in Neural Network Language Models
• Multimodal Word Distributions

Cited by in corpus (4)

• Natural Language Processing Advancements By Deep Learning: A Survey
• Position-Aware Self-Attention based Neural Sequence Labeling
• Data Augmentation for Neural Online Chat Response Selection
• How Does Adversarial Fine-Tuning Benefit BERT?