Skip to main content
. 2019 Oct 30;10(11):862. doi: 10.3390/genes10110862

Figure 1.

Figure 1

The detailed layouts of the three types of convolutional neural networks (ConvNet1, ConvNet2, and ConvNet3). ⊕ denotes element-wise addition. The dashed box in ConvNet1 highlights a local residual learning block. The blue lines in ConvNet1 denote hierarchical features. The “Concat” layer in ConvNet1 is used to concatenate all hierarchical features from each of the local residual learning blocks. HiCNN2-1 only uses ConvNet1; HiCNN2-2 is an ensemble of ConvNet1 and ConvNet2; and HiCNN2-3 is an ensemble of the three ConvNets.