Debiasing Multimodal Sarcasm Detection with Contrastive Learning

TL;DR

This paper introduces a contrastive learning framework for multimodal sarcasm detection that reduces textual bias and improves out-of-distribution generalization by using counterfactual data augmentation and debiasing techniques.

Contribution

It proposes a novel contrastive learning approach with counterfactual data augmentation to mitigate textual bias in multimodal sarcasm detection, enhancing OOD robustness.

Findings

The proposed framework outperforms existing methods in OOD sarcasm detection.

Counterfactual data augmentation effectively reduces textual bias.

Debiasing contrastive learning improves model robustness and generalization.

Abstract

Despite commendable achievements made by existing work, prevailing multimodal sarcasm detection studies rely more on textual content over visual information. It unavoidably induces spurious correlations between textual words and labels, thereby significantly hindering the models' generalization capability. To address this problem, we define the task of out-of-distribution (OOD) multimodal sarcasm detection, which aims to evaluate models' generalizability when the word distribution is different in training and testing settings. Moreover, we propose a novel debiasing multimodal sarcasm detection framework with contrastive learning, which aims to mitigate the harmful effect of biased textual factors for robust OOD generalization. In particular, we first design counterfactual data augmentation to construct the positive samples with dissimilar word biases and negative samples with similar word biases. Subsequently, we devise an adapted debiasing contrastive learning mechanism to empower the model to learn robust task-relevant features and alleviate the adverse effect of biased words. Extensive experiments show the superiority of the proposed framework.