Table 2 Classification accuracy for low vs high arousal/valence using DWT-based feature extraction using 32 EEG channels, proven with three classifiers and different signal segment sizes.

From: Two-dimensional CNN-based distinction of human emotions from EEG channels selected by multi-objective evolutionary algorithm

Segment (s)

Low vs. high arousal

Low vs. high valence

SVM

NB

kNN

SVM

NB

kNN

60

0.658 ± 0.23

0.629 ± 0.25

0.687 ± 0.23

0.571 ± 0.27

0.577 ± 0.28

0.645 ± 0.25

10

0.638 ± 0.15

0.605 ± 0.16

0.629 ± 0.13

0.614 ± 0.15

0.574 ± 0.17

0.605 ± 0.15

5

0.621 ± 0.13

0.599 ± 0.14

0.606 ± 0.19

0.608 ± 0.13

0.568 ± 0.15

0.600 ± 0.12

2

0.605 ± 0.11

0.586 ± 0.12

0.590 ± 0.09

0.584 ± 0.11

0.561 ± 0.13

0.574 ± 0.10

Back to article page