Correlation-Decoupled Knowledge Distillation for Multimodal Sentiment Analysis with Incomplete Modalities

TL;DR

This paper introduces a novel framework for multimodal sentiment analysis that effectively handles incomplete modalities by decoupling correlations and employing specialized knowledge distillation techniques, leading to improved performance.

Contribution

The paper proposes a correlation-decoupled knowledge distillation framework with sample-level contrastive, category-guided prototype, and response-disentangled strategies for robust multimodal sentiment analysis with missing data.

Findings

Achieves significant performance improvements over baselines.

Effectively handles incomplete modalities in real-world scenarios.

Demonstrates robustness across three datasets.

Abstract

Multimodal sentiment analysis (MSA) aims to understand human sentiment through multimodal data. Most MSA efforts are based on the assumption of modality completeness. However, in real-world applications, some practical factors cause uncertain modality missingness, which drastically degrades the model's performance. To this end, we propose a Correlation-decoupled Knowledge Distillation (CorrKD) framework for the MSA task under uncertain missing modalities. Specifically, we present a sample-level contrastive distillation mechanism that transfers comprehensive knowledge containing cross-sample correlations to reconstruct missing semantics. Moreover, a category-guided prototype distillation mechanism is introduced to capture cross-category correlations using category prototypes to align feature distributions and generate favorable joint representations. Eventually, we design a response-disentangled consistency distillation strategy to optimize the sentiment decision boundaries of the student network through response disentanglement and mutual information maximization. Comprehensive experiments on three datasets indicate that our framework can achieve favorable improvements compared with several baselines.