Pre-Training CNNs Using Convolutional Autoencoders

Despite convolutional neural networks being the state of the art in almost all computer vision tasks, their training remains a difficult task. Unsupervised representation learning using a convolutional autoencoder can be used to initialize network weights and has been shown to improve test accuracy after training. We reproduce previous results using this approach and successfully apply it to the difficult Extended Cohn-Kanade dataset for which labels are extremely sparse but additional unlabeled data is available for unsupervised use.