Fig. 5
From: Deep learning for DNase I hypersensitive sites identification
The figure shows that the overall construction of the model, which including 5 convolution layers, 4 maximum pooling layers, 1 SPP layer, and 3 fully connected layers. In addition to these visible structures, every layer is activated by LeakyReLU and followed by the dropout layer (the parameter is 0.3), and each fully connected layer is normalized by the batch normalization (BN, whose parameter is 0.5), which can speed up the convergence of the network. Dropout layers and BN layers are not depicted in the diagram