Cancer is the leading cause of death in Canada, and early diagnostics significantly increase the chances of treatment and survival. However, this process is tedious and often leads to a disagreement between pathologists. Computer-aided diagnosis (CAD) systems have shown improving diagnostic accuracy. However, researchers need to further study to improve the accuracy of CAD tools. This study develops a computational approach for colon cancer classification based on deep convolution neural networks (CNN). I use hematoxylin and eosin-stained dataset collected from the Chaoyang Hospital. This approach contains two deep CNN architectures from the computer vision domain for representational learning of cancer classes. I compare the results of the two CNN architectures, ResNet50 and ResNeXt50, and their performance on the Chaoyang dataset. By training, the ResNeXt achieves the state-of-the-art result of 86.21%.