Semi-supervised Classification Using Deep Learning
Bruno Vicente Alves de Lima, Adrião Duarte Dória Neto, Lúcia Emília Soares Silva, Vinícius Ponte Machado, Joao Guilherme Cavalcanti Costa · 2019
Semi-supervised learning considers the problem of classification when only a small subset of the observations have corresponding classes labels. Unfortunately, labeling a sufficiently large training data set will require human involvement, an expensive and time-consuming task. Recently, several papers in literature show that deep learning techniques are able to solve this problem. In this sense, this work proposes a Semi-supervised Deep Classification (SDC) method to solve the semi-supervised problem by using a Multilayer Perceptron (MLP) embedded to the labeling process. Extensive experiments on benchmark data sets validate the effectiveness and robustness of the proposed method.