Fig. 1From: A sparse autoencoder-based deep neural network for protein solvent accessibility and contact number predictionIllustration of Stacked Sparse Autoen-coder (SSAE) by three hidden layersBack to article page