A Novel Approach to Train Diverse Types of Language Models for Health Mention Classification of Tweets

TL;DR

This paper introduces an adversarial training method with Gaussian noise and contrastive loss to improve health mention classification in tweets, addressing challenges from figurative language and non-health uses.

Contribution

It proposes a novel adversarial training approach with layer-specific noise addition and contrastive loss for better health mention classification in social media texts.

Findings

Adding noise at earlier layers improves performance.

Adding noise at intermediate layers deteriorates performance.

Final layer noise addition yields the best results.

Abstract

Health mention classification deals with the disease detection in a given text containing disease words. However, non-health and figurative use of disease words adds challenges to the task. Recently, adversarial training acting as a means of regularization has gained popularity in many NLP tasks. In this paper, we propose a novel approach to train language models for health mention classification of tweets that involves adversarial training. We generate adversarial examples by adding perturbation to the representations of transformer models for tweet examples at various levels using Gaussian noise. Further, we employ contrastive loss as an additional objective function. We evaluate the proposed method on the PHM2017 dataset extended version. Results show that our proposed approach improves the performance of classifier significantly over the baseline methods. Moreover, our analysis shows that adding noise at earlier layers improves models' performance whereas adding noise at intermediate layers deteriorates models' performance. Finally, adding noise towards the final layers performs better than the middle layers noise addition.