PairCFR: Enhancing Model Training on Paired Counterfactually Augmented Data through Contrastive Learning

TL;DR

PairCFR introduces a contrastive learning approach to improve model training on counterfactually augmented data, enhancing robustness and reducing bias by promoting broader feature utilization.

Contribution

This work is the first to integrate contrastive learning with counterfactually augmented data to mitigate bias and improve out-of-distribution performance.

Findings

Outperforms state-of-the-art on OOD datasets

Encourages models to use a wider range of features

Theoretically proves contrastive loss promotes broader feature use

Abstract

Counterfactually Augmented Data (CAD) involves creating new data samples by applying minimal yet sufficient modifications to flip the label of existing data samples to other classes. Training with CAD enhances model robustness against spurious features that happen to correlate with labels by spreading the casual relationships across different classes. Yet, recent research reveals that training with CAD may lead models to overly focus on modified features while ignoring other important contextual information, inadvertently introducing biases that may impair performance on out-ofdistribution (OOD) datasets. To mitigate this issue, we employ contrastive learning to promote global feature alignment in addition to learning counterfactual clues. We theoretically prove that contrastive loss can encourage models to leverage a broader range of features beyond those modified ones. Comprehensive experiments on two human-edited CAD datasets demonstrate that our proposed method outperforms the state-of-the-art on OOD datasets.