. 2021 May 14;21(10):3414. doi: 10.3390/s21103414

Table 1.

A brief summary of the analyzed works and their main characteristics.

Work	Database	Features (Brain Waves)	Classifier	Emotions (#Classes)
[17]	MANHOB-HCI	PSD, DPSA ( $θ, α, β, γ$ )	SVM	arousal (3); valence (3)
[2]	DEAP	PSD, APSD ( $θ, α, β, γ$ )	NB	arousal (2); valence (2)
[41]	Video (Own)	SampEn, Spectral Centroid ( $α, β, γ$ )	KNN, PNN	disgust, happy, surprise, fear and neutral (5)
[18]	DEAP	PSD ( $θ, α, β, γ$ )	DLN, SVM, NB	arousal (3); valence (3)
[15]	Video (Own)	DPSA, WT, WE, AE, FD, HE ( $δ, θ, α, β, γ$ )	SVM	positive and negative (2)
[19]	DEAP	Pearson correlation, Phase coherence, MI ( $θ, α, β, γ$ )	SVM	arousal (2); valence (2)
[16]	SEED	PSD, DE, Differential/Rational asymmetry ( $δ, θ, α, β, γ$ )	KNN, LR, SVM, DBN	positive, negative and neutral (3)
[27]	DEAP	PSD, STFT, HHS, HOC ( $θ, α, β, γ$ )	RF, SVM	anger, surprise, other (3)
[29]	DEAP	Statistical, PSD, HP, FD ( $θ, α, β, γ$ )	SVM	arousal (2); valence (2)
[31]	DEAP	WP, WE ( $θ, α, β, γ$ )	SVM, KNN	arousal (2); valence (2)
[43]	DEAP, Music	PSD, FD, differential asymmetry ( $δ, θ, α, β, γ$ )	SVM, MLP, C4.5	arousal (2); valence (2)
[13]	MANHOB-HCI	EMD, SampEn ( $β$ )	SVM	high/low valence/arousal (4)
[14]	DEAP	EMD, AR ( $β$ )	SVM	high/low valence/arousal (4)
[4]	DREAMER	Logarithmic PSD ( $θ, α, β$ )	SVM	arousal (2); valence (2)
[3]	AMIGOS	Logarithmic PSD, APSD ( $θ, α, β, γ$ )	NB	arousal (2); valence (2)
[21]	DEAP	WP ( $δ, θ, α, β, γ$ )	ANN, SVM	high/low valence/arousal (4)
[44]	DEAP	Quadratic time-frequency distributions (Custom)	SVM	high/low valence/arousal (4)
[23]	DEAP, SEED	Flexible Analytic WT, Rényi’s Quadratic Entropy (Custom)	SVM, RF	high/low valence/arousal (4); positive, negative and neutral (3)
[42]	DEAP	PSD, APSD, Shannon Entropy, SE, ZCR, Statistical ( $θ, α, β, γ$ )	LSSVM (Least Square SVM)	joy, peace, anger and depression (4)
[45]	DEAP	WP, WE ( $θ, α, β, γ$ )	KELM	high/low valence/arousal (4)
[20]	DEAP, SEED	WT, High Order Statistics ( $δ, θ, α, β, γ$ )	DLN	high/low valence/arousal (4); positive, negative and neutral (3)
[22]	Video (Own)	LZC, WT, Cointegration Degree, EMD, AE (Custom)	SVM	arousal (2); valence (2)

Feature Extraction: AE—Approximate Entropy, APSD—Asymmetric Power Spectrum Density, AR—Auto Regressive models, DE— Differential Entropy, DPSA—Differential Power Spectral Asymmetry, EMD—Empirical Mode Decomposition, FD—Fractal Dimensions, HE—Hurst Exponent, HHS—Hilbert–Huang Spectrum, HOC—Higher Order Crossings, HP—Hjorth Parameters, LZC—Lempel-Ziv Complexity, MI—Mutual Information, PSD—Power Spectrum Density, SampEn—Sample Entropy, SE—Spectral Entropy, STFT—Short- Time Fourier Transform, WE—Wavelet Entropy, WP—Wavelet Energy, WT—Wavelet Transform and ZCR—Zero-Crossing Rate. Classifier: ANN—Artificial Neural Networks, DBN—Deep Belief Networks, DLN—Deep Learning Networks, KELM—Extreme Learning Machine with kernel, KNN—K-Nearest Neighbors, LR—Logistic Regression, MLP—Multi-Layer Perceptron, NB—Naive Bayes, RF—Random Forest and SVM—Support Vector Machines.