Skip to content

Files

Latest commit

Aug 23, 2017
5cd9bf7 · Aug 23, 2017

History

History
16 lines (9 loc) · 578 Bytes

README.md

File metadata and controls

16 lines (9 loc) · 578 Bytes

keras odds and ends

Stacked what where autoencoders in tensorflow and keras.

mnist_swwae-tf.py, tensorflow compatible version of mnist_swwae.py from https://github.com/fchollet/keras/blob/master/examples/mnist_swwae.py

mnist_swwae-pure-tf.py, same as above but uses pure tensorflow, no keras.

mnist_swwae-pure-tf-paper.py, uses pure tensorflow but implements L2M error terms not computed in the above better matching derivation in paper [1].

[1] Stacked What-Where Auto-encoders Junbo Zhao, Michael Mathieu, Ross Goroshin, Yann LeCun https://arxiv.org/abs/1506.02351