Reliability-aware disentangled adaptive network for multimodal sentiment analysis

DiMoE: Disentangled Representation Learning with Mixture of Experts Fusion for Sentiment Intensity Prediction and Emotion Classification

DiMoE is proposed, a representation-first framework that integrates feature disentanglement with sparse Mixture of Experts (MoE) fusion, indicating that explicit modality disentanglement combined with Mixture of Experts fusion yields effective multimodal affective representations.

Reliability-aware Cross-sample Enhancement for Robust Multimodal Sentiment Analysis

This work introduces an adaptive variational information bottleneck to model modality-wise uncertainty and perform quality-aware information compression, thereby suppressing redundant noise in unreliable modalities and designs a reliability-aware cross-sample enhancement strategy that retrieves high-confidence, semanti...

Dynamic graph interaction for feature-augmented multimodal sentiment analysis

Register-augmented attention and self-calibrated fusion for robust multimodal sentiment analysis

Results indicate that combining register-based representation stabilization with quality-aware fusion provides an effective solution for robust multimodal sentiment analysis, and RegCal-Net achieves competitive performance across multiple evaluation metrics, particularly improving fine-grained sentiment classification...

Label-Aware Entropic Distributional Contrastive Alignment for Multimodal Sentiment Analysis

Multimodal sentiment analysis integrates text, audio, and visual signals to infer affective states. However, sentiment information shared across modalities is often entangled with modality-specific variation, and existing representation learning methods do not fully exploit the relations encoded by continuous sentiment...

SemMSA: Latent Semantic-Aided Robust Multimodal Sentiment Analysis with Incomplete Data

SemMSA, a latent semantic-aided framework that constructs rich sentiment-relevant semantics with LLMs, fully integrating with all modalities via anchor-free spectral alignment, is proposed.