Table 4.

Comparison between the proposed approach and the existing methods

Comparison Parameters	Proposed Approach	Hill [29]	Li et al. [45]	Ma et al. [46]	Aneja et al. [33]	Aneja et al. [47]
Number of emotions recognized	4 (H, S, A, & Su)	3 (H, S & A)	4 (A, H, S, & N)	6 (H, S, A, F, Su, & D)	7 (S, A, Su, D, F, N, & J)	7 (S, A, Su, D, F, N, & J)
Learning rate	0.0003	0.001	–	–	0.01	0.0001
Accuracy Score for Individual Emotion	0.97 (H), 0.95 (S), 0.96 (A), 0.96 (Su)	–	0.77 (A), 0.65 (H), 0.70 (S)	0.78 (H), 0.49 (S), 0.62 (A), 0.19 (Su)	0.89 (S), 0.85 (A), 0.95 (Su)	0.79 (S), 0.90 (A), 0.94 (Su)
Overall Model Accuracy Score	0.96	0.80	–	–	0.89	–
F1-score	0.85	–	–	–	–	–

^*H → Happy, S → Sad, A → Angry, Su → Surprised, F → Fear, D → Disgusted, J → Joy, N→ Neutral