EEG-based classification of visually evoked emotions using multiscale entropy features

Analysing the indication of emotion in electroencephalogram (EEG) signals develops the accuracy of existing emotion recognition techniques. Therefore, Audiovisual emotion real-live recognition is still challenging in real-life data acquisition cases, including changing and challenging surrounding conditions and indoor and outdoor scenarios [1]. Multiscale entropy measures have been extensively introduced from the beginning of the last century to evaluate the complexity of time-domain in physical systems. Since then, these approaches have been used in a broad spectrum of applications and have got considerable attention. This paper introduced a multiscale information analysis of EEG signals for distinguishing emotional states in four dimensions based on entropy features. The algorithms were applied to extract features on SJTU Emotion EEG Dataset (SEED), which included multiscale EEG complexity index in the time domain, and ensemble empirical mode decomposition enhanced energy and fuzzy entropy. In covered studies, researchers used emotional videos to evoke emotion, and then the corresponding signals were collected [2]. Here we are using the features from different brainwaves, frequency, and entropy to identify the subjects' happiness status. The entropy computes the original signal and the time domain to evaluate the irregularity for each node. Moreover, different entropy measures are shown superior to other entropy features such as approximate entropy, sample entropy, permutation entropy, fuzzy entropy, and distribution dispersion entropy.