This is a personal attempt to reimplement a contractive autoencoder (with FC layers uniquely) as described in the original paper by Rifai et Al.

To the best of our knowledge, this is the first implementation done with native Tensorflow.

I also provide in this repository extensions to the original contractive loss, to include a contraction term I computed for (in addition to the one provided in the paper for 1 layer of sigmoid non linearity):

• 2 layers of sigmoid non linearity
• 1 layer of relu non linearity
• 2 layers of relu non linearity

More details here.

[1] Salah Rifai, Pascal Vincent, Xavier Muller, Xavier Glorot, and Yoshua Bengio. Contractive auto-encoders: Explicit invariance during feature extraction. In Proceedings of the 28th International Conference on International Conference on Machine Learning, ICML’11, pages 833–840, USA, 2011. Omnipress.