Infant asphyxia detection using autoencoders trained on locally linear embedded-reduced Mel Frequency Cepstrum Coefficient (MFCC) features

Deep-learning methods are representation, obtained by composing simple but non representation at one level (starting with the raw input) into a representation at a higher, slightly more abstract level.With the composition of enough such transformations, very complex functions can be learned and solved.The objective of this paper is to explore one of the Deep Learning paradigms called autoencoders to perform diagnosis of infant asphyxia.LLE was used to reduce size of feature representation while enhancing them.Two stacked autoencoders were then trained to extract the necessary features for clasification.Extensive tests performed showed that the best classification accuracy.