Multi-Modal Classification Model with Attention-Based Fusion for Enhanced Accuracy

In response to the inefficiency of the vector concatenation-based multi-modal classification model and the neglect of single-modal information in multi-modal classification models, we propose a new multi-modal classification model that employs a multi-modal information fusion method based on attention mechanism. This model can effectively improve classification accuracy and utilize single-modal information for auxiliary classification. The model demonstrates good performance in experiments, proving its feasibility and effectiveness in multi-modal classification tasks.