The field of multimodal emotion recognition and sentiment analysis is moving towards the development of more sophisticated and effective models that can handle the complexities of multimodal data and class imbalance. Researchers are exploring new architectures and techniques, such as cross-attention networks, contrastive learning, and hybrid deep neural networks, to improve the accuracy and robustness of emotion recognition and sentiment analysis systems. These innovations are addressing key challenges, including modal heterogeneity, category imbalance, and contextual nuances, and are leading to significant improvements in performance. Notable papers in this area include: MCN-CL, which proposes a multimodal cross-attention network and contrastive learning approach that outperforms state-of-the-art methods on benchmark datasets. TiCAL, which introduces a typicality-based consistency-aware learning framework that mitigates inter-modal emotion conflicts and enhances overall recognition accuracy.