Attention-Enhanced Multimodal Sentiment Analysis Using Resnet50-Cbam, BERT, And Graph Neural Networks
The present study proposes such a multimodal sentiment analysis framework with attention-enhanced properties, a combination of ResNet50 and Convolutional Block Attention Module (CBAM), a textual encoder with BERT, and refinement of relational features via Graph Neural Networks (GNN). The model is designed to address th...