Investigating Ensemble Methods for Model Robustness Improvement of Text Classifiers

TL;DR

This paper explores how ensemble methods can enhance the robustness of text classifiers by selecting appropriate bias models, addressing the limitations of fixed low-capacity models in handling diverse bias features.

Contribution

It introduces an analysis of bias features and demonstrates that choosing suitable bias models improves robustness over fixed low-capacity models.

Findings

No single bias model works best for all cases

Appropriate bias model selection leads to better robustness

Ensemble methods outperform fixed models in robustness

Abstract

Large pre-trained language models have shown remarkable performance over the past few years. These models, however, sometimes learn superficial features from the dataset and cannot generalize to the distributions that are dissimilar to the training scenario. There have been several approaches proposed to reduce model's reliance on these bias features which can improve model robustness in the out-of-distribution setting. However, existing methods usually use a fixed low-capacity model to deal with various bias features, which ignore the learnability of those features. In this paper, we analyze a set of existing bias features and demonstrate there is no single model that works best for all the cases. We further show that by choosing an appropriate bias model, we can obtain a better robustness result than baselines with a more sophisticated model design.