Natural Language Inference

Attention Meets Perturbations: Robust and Interpretable Attention with Adversarial Training

IEEE Access (**Impact Factor: 3.367** in 2020; [1st place in Engineering & Computer Science (general) at Google Scholar Metrics](https://scholar.google.com/citations?view_op=top_venues&hl=en&vq=eng_enggeneral))

Making Attention Mechanisms More Robust and Interpretable with Virtual Adversarial Training for Semi-Supervised Text Classification

We propose a new general training technique for attention mechanisms based on virtual adversarial training (VAT). VAT can compute adversarial perturbations from unlabeled data in a semi-supervised setting for the attention mechanisms that have been …