Skip to main content
. 2019 Apr 9;35(21):4222–4228. doi: 10.1093/bioinformatics/btz251

Fig. 1.

Fig. 1.

The architecture of our convolutional neural network. There are five types of layers, including conv1, conv2, conv3, conv4R and conv5. Edge ‘1’ marks the global residual learning, and Edge ‘2’ marks the local residual learning. The dashed box highlights a building block for local residual learning. ⊕ denotes element-wise addition