Fusing highly dimensional energy and connectivity features to identify affective states from EEG signals

Abstract

In this paper, a novel method for affect detection is presented. The method combines both connectivity-based and channel-based features with a selection method that considerably reduces the dimensionality of the data and allows for an efficient classification. In particular, the Relative Energy (RE) and its logarithm in the spacial domain, and the spectral power (SP) in the frequency domain are computed for the four typical frequency bands (α, β, γ and θ), and complemented with the mutual information measured over all channel pairs. The resulting features are then reduced by using a hybrid method that combines supervised and unsupervised feature selection. First, a Welch's t-test is used to select the features that best separate the classes, and discard the ones that are less useful for classification. To this end, all features where the t-test yields a p-value above a threshold are eliminated. The remaining ones are further reduced by using Principal Component Analysis. Detection results are compared to state-of-the-art methods on DEAP, a database for emotion analysis composed of labeled recordings from 32 subjects while watching 40 music videos. The effect of using different classifiers is also evaluated, and a significant improvement is observed in all cases.

On the Influence of Affect in EEG-Based Subject Identification

“…EEG signals have been commonly used in medicine to diagnose a diversity of pathological conditions and disorders [9], [10], [11], [12], [13], but their use has recently been extended to other fields of research. At the beginning of this century, EEG correlates of emotions reported in a number of neuropsychological studies [14] also motivated their use in the emotion (affect) recognition field [15], [16], [17], [18]. In emotion recognition, feature selection has been proved to significantly affect the classification performance [14], [19].…”

“…In order to unify the experiments across all different databases and yet maintain consistency with other previous works in the literature that use DEAP and MANHOB, e.g. [16], [36], [41], [42], the arousal and valence labels in these two datasets were discretized into two levels by applying a threshold to the originally reported values. The threshold was set to 5, in order to be coherent with other previous studies [16], [36], [41], [42].…”

“…[16], [36], [41], [42], the arousal and valence labels in these two datasets were discretized into two levels by applying a threshold to the originally reported values. The threshold was set to 5, in order to be coherent with other previous studies [16], [36], [41], [42]. For consistency with the DEAP and MAHNOB datasets, the samples of the SEED dataset that referred to neutral emotion were discarded, and only those referring to positive or negative emotion were used in this study.…”

Emotion detection from EEG recordings based on supervised and unsupervised dimension reduction

“…Meanwhile, Gupta et al also applied connectivity features for the DEAP dataset . This method is inspired by the work of Pablo et al For their work, Welchs t‐test and PCA used the two‐step feature reduction methods, which applied for the fusion of spectral power and mutual information. At the DEAP dataset, a total of 880 features were extracted from 32 channels.…”

Computational model for detection of abnormal brain connections in children with autism

“…The relative energy quantifies signal strength as it gives the area under the curve as the power within any temporal interval. In signal processing, the energy of a finite EEG signal is given by (4) [38,39,43]:…”