Lately, research on computational models of emotion had been getting much attention due to their potential for understanding the mechanisms of emotions and their promising broad range of applications that potentially bridge the gap between human and machine interactions. We propose a new method for emotion classification that relies on features extracted from those active brain areas that are most likely related to emotions. To this end, we carry out the selection of spatially compact regions of interest that are computed using the brain neural activity reconstructed from Electroencephalography data. Throughout this study, we consider three representative feature extraction methods widely applied to emotion detection tasks, including Power spectral density, Wavelet, and Hjorth parameters. Further feature selection is carried out using principal component analysis. For validation purpose, these features are used to feed a support vector machine classifier that is trained under the leave-one-out cross-validation strategy. Obtained results on real affective data show that incorporation of the proposed training method in combination with the enhanced spatial resolution provided by the source estimation allows improving the performed accuracy of discrimination in most of the considered emotions, namely: dominance, valence, and liking.

No abstract

The estimation of current distributions from electroencephalographic recordings poses an inverse problem, which can approximately be solved by including dynamical models as spatio-temporal constraints onto the solution. In this paper, we consider the electrocardiography source localization task, where a specific structure for the dynamical model of current distribution is directly obtained from the data by fitting multivariate autoregressive models to electroencephalographic time series. Whereas previous approaches consider an approximation of the internal connectivity of the sources, the proposed methodology takes into account a realistic structure of the model estimated from the data, such that it becomes possible to obtain improved inverse solutions. The performance of the new method is demonstrated by application to simulated electroencephalographic data over several signal to noise ratios, where the source localization task is evaluated by using the localization error and the data fit error. Finally, it is shown that estimating MVAR models makes possible to obtain inverse solutions of considerably improved quality, as compared to the usual instantaneous inverse solutions, even if the regularized inverse of Tikhonov is used.

En este artículo se presenta un método de estimación de la actividad neuronal sobre el cerebro usando un filtro de Kalman con suavizado, que tiene en cuenta en la solución del problema inverso, la variabilidad dinámica de la serie de tiempo. Este método es aplicado sobre un modelo realista de la cabeza, calculado con elementos finitos de frontera. Se presenta un análisis comparativo entre diferentes métodos de estimación y el método propuesto sobre señales EEG simuladas para diferentes condiciones de relación señal a ruido. La solución del problema inverso se hace utilizando computación de alto desempeño y se presenta una evaluación delcosto computacional para cada método. Como resultado, el filtro de Kalman con suavizado presenta un mejor desempeño en la tarea de estimación comparado con la solución estática regularizada, y la solución dinámica sin suavizado.

No abstract