Large-scale data generation has brought the need for the development
of intelligent techniques capable of analyzing this data automatically. In this
sense, this paper proposes a semisupervisioned classification model capable of
labeling unlabeled data from a few labeled examples. For this, a deep neural
network was trained with labeled and unlabeled examples, simutaneally. The
experiments performed show that the model is efficient in labeling data and
predicting new examples.