Classificação de emoções faciais utilizando a rede neural sem pesos WiSARD

Abstract

Automatic classification of emotions in facial expressions is a central issue in Affective Computing and one of the main premises in the construction of increasingly responsive man-machine interface models with a wide range of applications. Many systems based on artificial intelligence are able to solve this problem with high accuracy, but in general such models have a slow and expensive learning process. The recognition of facial expressions through the use of a WiSARD-based n-tuple classifier is explored in this work. The competitiveness of this weightless neural network is tested in the specific challenge of identifying emotions from photos of faces, limited to the six basic emotions described in the seminal work of Ekman and Friesen (1977) on identification of facial expressions. Experiments carried out with the two main datasets found in the literature demonstrated their competitiveness with current state-of-the-art, as well as their great speed in both the learning and classification phases. Different preprocessing approaches as well as studies on how best to represent images in binary inputs in this specific problem are also described in this text.