Stress Classification Using Brain Signals Based on LSTM Network

Table 4

Classification accuracy comparison for stress classification. The table shows maximum (Max), average (avg), and minimum (Min) stress classification accuracy obtained with different methods.

Method	Validation method	Accuracy
		Max	Avg	Min
MLP	50–50	70.11	69.33	68.56
LSTM 1	50–50	83.33	81.73	80.14
LSTM 2	50–50	87.22	85.79	84.36
MLP	60–40	73.48	71.80	70.12
LSTM 1	60–40	84.69	82.75	80.76
LSTM 2	60–40	89.28	88.22	87.16
MLP	70–30	75.71	73.77	71.84
LSTM 1	70–30	85.82	83.14	81.87
LSTM 2	70–30	90.33	91.11	91.89
MLP	10-fold cross validation	76.27	72.01	70.59
LSTM 1	10-fold cross validation	87.61	86.55	85.97
LSTM 2	10-fold cross validation	93.17	91.96	90.76