基于多视角解耦融合的多模态会话情感识别

张婷婷,朱小飞

Multi-view decoupled fusion for multimodal emotion recognition in conversation

Tingting ZHANG,Xiaofei ZHU

表 4 不同模态组合输入在2个数据集上的消融实验结果

Tab.4 Ablation results of different modal combination inputs on two datasets

模型 IEMOCAP MELD Acc/% W-F1/% Acc/% W-F1/% A 53.10 50.36 46.88 41.49 V 39.90 38.99 45.70 35.52 T 69.03 69.03 67.41 66.40 A+T 70.88 70.82 67.79 66.54 V+T 68.66 68.87 67.35 66.37 A+V 64.31 63.86 49.20 43.02 MDF 72.87 73.05 68.06 66.92