Figure 6.

Construction of the prediction model based on the convolutional neural network and convolutional autoencoder for learning the attention representation and the low-dimensional network representation.