Discriminative stacked autoencoder for feature representation and classification

摘要

＜正＞Dear editor,Recently, deep learning （DL） has become a hot research topic and as one of the most well-known DL models, stacked autoencoder （SAE）[1] has received increasing attention. In SAE, layer-wise pretraining is the basic mechanism for automatic feature extraction and it can also avoid gradient vanishing while constructing deep architectures.