DCAP: A deep convolution architecture for prediction of urban growth

摘要

This work proposes a new deep convolution architechture for predicting urban growth using satellite images. The case study shows that even without any feature extraction algorithm, performance of DCAP is comparable to MLP and RBFN. Further, other kinds of deep architechtures, such as deep belief net or recurrent neural net, can be applied and their performance can be compared.