Table 7.

Comparison of emotion recognition studies from EEG signals

References (year)	Database	Method	Number of classes	Accuracy (%)
Atkinson and Campos (2016)	DEAP	Statistical features, FD, HP, mRMR, SVM	2	73.41 (valence), 73.06 (arousal)
Yang et al. (2018)	DEAP	Empirical mode decomposition, Sample entropy, support vector machine	4	93.2
Zheng et al. (2017)	DEAP	Differential entropy, Graph regularized Extreme Learning Machine	4	69.67
Soroush et al. (2018)	DEAP	Nonlinear features (CD, FD, …), ICAs, modified Dempster-Shafer theory of evidence	4	90.54
Soroush et al. (2020)	DEAP	Poincare plane, MSVM, KNN, MLP	4	89.76
Lee and Hsieh (2014)	DEAP	PSI, Coh, Corr, SVM, MLP, DT	2	73.30 (arousal), 72.50 (valence)
Zhang et al. (2017)	DEAP	MI, PCC, SVM, RF	2	72.6 (valence), 70.3 (arousal)
Li et al. (2019)	DEAP, MAHNOB-HCI	PLV, SVM, GELM	4	68 (MAHNOB-HCI), 62 (DEAP)
Wang et al. (2019)	DEAP	PLV, graph-CNN	2	73.31 (valence), 77.03 (arousal)
Yang et al. (2018)	DEAP	Recurrence quantification analysis, Parallel Convolutional Recurrent Neural Network	2	90.8 (valence), 91.03 (arousal)
Xiaofen et al. (2019)	DEAP	Stack AutoEncoder-Long short-term memory	2	81.10 (valence), 74.38 (arousal)
Yang et al. (2019)	DEAP	Multi column CNN	2	90.01 (valence), 90.65 (arousal)
Shen et al. (2020)	DEAP	Differential entropy, 4D-Convolutional recurrent neural network	2	94.22 (valence), 94.58 (arousal)
Proposed method	DEAP, MAHNOB-HCI	dDTF, PDC, ResNet-50, Inception-v3, AlexNet, VGG-19	5	98.16 ± 0.54 (dDTF, DEAP), 99.43 ± 0.58 (dDTF, MAHNOB-HCI)