Zongheng Wang

Multi-Metric Representation Learning Strategy Based on Clustering for Fine-Grained Multimodal Sentiment Analysis

Multimodal sentiment analysis (MSA) aims to identify human emotions through multimodal data. Despite considerable advances in MSA, we find that emotional class centers often overlap when integrating data from different modalities into the same representation space. In this paper, we propose a novel