Table 5.

The performance of our models using visual, audio, and text features.

Models (Video, Audio, and Subtitle)	Extended COGNIMUSE (Intended Emotion)
	Arousal		Valence
	MSE	PCC	MSE	PCC
Feature AAN	0.117	0.655	0.170	0.575
Temporal AAN	0.149	0.560	0.226	0.387
Mixed AAN	0.198	0.310	0.267	0.275
2FC-layer model	0.289	0.229	0.283	0.227
2-layer LSTM model	0.223	0.080	0.277	0.119

Movie subtitles are not available in the Global EIMT16 dataset.