Unsupervised learning of non-Gaussian mixtures with temporal dependencies

Classification methods typically make use only of labeled data, in what is known as supervised learning. In some applications, however, labeled data is either scarce or costly to obtain. For these applications, unsupervised or semisupervised learning are adequate, since they will be able to use unlabeled data. This work proposes a new method for unsupervised and semisupervised learning of non-Gaussian data mixtures with temporal dependencies represented by sequential independent component analysis mixture models (SICAMM). The proposed method was applied on simulated and real data, and its classification performance was compared with that of supervised learning SICAMM and ICAMM, and two semisupervised learning Bayesian networks. The real data application consisted of the detection of microarousals in sleep electroencephalographic (EEG) recordings for the purposes of sleep disorder diagnosis. Results show that the proposed method obtained better performance that the other considered methods.