Emotion Analysis of College Students Using a Fuzzy Support Vector Machine

Table 1

Representative research on emotion recognition based on EEG signals.

Method	Feature	Representative research	Recognition rate (%)
Linear analysis methods (Pearson correlation, amplitude squared coherence, autoregressive model, cumulative energy algorithm, time-frequency analysis, etc.)	EEG signal waveform characteristics (such as amplitude, phase, etc.), rhythm wave average power, power spectral density, band energy, wavelet coefficient root mean square, etc.	References [6]	60.42
		References [7]	62.50
		References [37]	88.51
Nonlinear analysis methods (mutual information [38], correlation dimension, Lempel–Ziv (LZ) complexity, recursive graph, and entropy analysis [39])	Entropy, fractal dimension, correlation dimension, CO complexity, LZ complexity, Hust index, maximum Lyapunov index, etc.	References [40]	80.40
		References [41]	92.50
		References [42]	86.65