Tensorflow implementation of the Ladder Network with convolutional Layers. The Ladder Network consists of an Encoder and Decoder with lateral skip connections that jointly optimize a supervised and unsupervised cost function, thus making it extremely suitable for semi-supervised learning. This repository can be further extended by adding for example recurrent layers. See more information about the Ladder Network here: https://arxiv.org/abs/1507.02672 To view the paper: https://www.researchgate.net/publication/329377534_Ladder_Networks_for_Semi-Supervised_Hyperspectral_Image_Classification

If you use this implementation, please acknowledge our work by citing the paper.