A EEG-based emotion recognition model with rhythm and time characteristics

Brain Informatics

Table 9 Comparison of results that use EEG signals of DEAP dataset for emotion recognition

Literature	Emotion category	Window’s length	Classification	The highest classification accuracy (Acc/%)
Rozgić et al. [20]	Arousal/2 Valence/2	1 s/2 s/4 s/8 s (1-s step length)	SVM KNN	68.4/2 76.9/2
Zhuang et al. [21]	Arousal/2 Valence/2	1 s (0.1-s step length)	SVR	68.4/2 76.9/2
Yoon et al. [3]	Arousal/2 Valence/2	2 s (1-s step length)	Bayesian based on sensor convergence	70.1/2 70.9/2
Hatamikia et al. [22]	Arousal/2	1 s	KNN, QDA, LDA	74.2/2 72.33/2
Hatamikia et al. [22]	Valence/2	1 s	KNN, QDA, LDA	74.2/2 72.33/2
Tripathi et al. [23]	Arousal/2 Valence/2	–	DNN	73.28/2 75.58/2
Li et al. [24]	(Arousal and Valence)/2	3 s	SAE, LSTM RNN	79.26/2
Kuai et al. [25]	Arousal/2 Valence/2	3 s	RSP-ERM	64/2 66.6/2
Our work	Arousal/2	< 1 s	RT-REM	69.1/2
Our work	Valence/2	< 1 s	RT-REM	62.12/2

SVR support vector regression, QDA quadratic discriminant analysis, LDA linear discriminant analysis, RSP-ERM emotional recognition model based on rhythm synchronization patterns, /2 binary classification