Noise-Assisted Multivariate Empirical Mode Decomposition Based Emotion Recognition

Pinar Ozel, Aydin Akan, Bulent Yilmaz
2018 Electrica  
Electroencephalography (EEG), introduced by Hans Berger in 1924, is a noninvasive method that utilizes electrical potential recordings on the scalp at different locations at microvolt level. A brain-computer interface (BCI), an innovation based on computer-assisted controls utilizing brain activity, depends on EEG signals and provides a means to discover a variety of uses ranging from bioengineering to neuro-prosthetics. These new advancements in human-computer interaction applications also
more » ... lize the transfusion of emotional states regarding information between the brain and the computer. Thus, in the literature, there are numerous studies on emotional state modeling [1-3]. Yet, the most commonly used two-dimensional or three-dimensional space is a circumplex model that shows the emotional state as continuous points. In the two-dimensional space, emotions are displayed by arousal-valence map and with respect to three-dimensional space, they are modelled as arousal-valence-dominance (VAD) map. In these models of emotions, emotional states are one of the qualities of physiological-neural aspects of emotions, which are isolated from each other, and are represented as a blend of these dimensions. Arousal is defined as the power or intensity of sensation (emotional arousal), valence is defined as the satisfaction or dissatisfaction grade (emotional valence), and dominance is defining as the power of controlling emotion internally (emotional dominance). For example, anger is shown as a combination of negative valence and high arousal [4] . Traditional time-frequency representation algorithms, such as short-time Fourier transform (STFT) and continuous wavelet transform (CWT), have been utilized to frequently examine the emotional state data [5] .Yet such methods restrict the representations in time-frequency space depending on the projection of data onto the fixed arrangement of the fundamental ABSTRACT Emotion state detection or emotion recognition cuts across different disciplines because of the many parameters that embrace the brain's complex neural structure, signal processing methods, and pattern recognition algorithms. Currently, in addition to classical time-frequency methods, emotional state data have been processed via data-driven methods such as empirical mode decomposition (EMD). Despite its various benefits, EMD has several drawbacks: it is intended for univariate data; it is prone to mode mixing; and the number of local extrema must be enough before the EMD process can begin. To overcome these problems, this study employs a multivariate EMD and its noise-assisted version in the emotional state classification of electroencephalogram signals.
doi:10.26650/electrica.2018.00998 fatcat:cxojqk6tczepxncitwjcmdqfqy