In many practical applications of machine vision, a small number of samples are labeled and therefore, classification accuracy is low. On the other hand, labeling by humans is a very time consuming process, which requires ...