Keras implementation of Deep Learning papers
Clone or download
Latest commit c327d8c Apr 1, 2017
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
LICENSE Initial commit Mar 27, 2017
README.md Update README.md Apr 1, 2017

README.md

Keras Deep Learning Paper Implementations

A curated list of implementations in keras.

It's a bit of a hassle to find implementation of most of the latest papers. Hopefully this allows anyone to get up and running with the state-of-the-art networks in little to no time.

We welcome your contributions!

If you have any paper/code suggestions, please feel free to edit and sumbit a pull request.


Imagenet Models

Unsupervised / Generative Models

  • Pix2Pix. Image-to-Image Translation with Conditional Adversarial Networks (2016), P. Isola et al. [pdf] [code]
  • Deepmind's wavenet (2016), Van den Oord et al. [pdf] [code]
  • Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network. (2016), C. Ledig et al. [pdf] [code]

Convolutional Models

  • XCeption (2016), F. Chollet. [pdf] [code]
  • Inception v3 (2015), C. Szegedy et al. [pdf] [code]
  • Image Super-Resolution Using Deep Convolutional Networks (2015), C. Dong et al. [pdf] [code]
  • Time-series modeling with undecimated fully convolutional neural networks (2015), R. Mittelman. [pdf] [code]
  • DenseNet: Densely Connected Convolutional Network (2016), G. Huang. [pdf] [code]

LSTM

  • Phased LSTM: Accelerating Recurrent Network Training for Long or Event-based Sequences (2016), D. Neil [pdf] [code]
  • Bidirectional LSTM: Neural Architectures for Named Entity Recognition (2016), G. Lample [pdf] [code]

Fun Models

  • Deep Dream. Inceptionism: Going Deeper into Neural Networks (2015), A. Mordvintsev et al. at Google. [blog] [code]
  • Style Transfer. Image Style Transfer Using Convolutional Neural Networks (2016), L. Gatys. [pdf] [code]
  • Fast (realtime) Neural Style Transfer. Perceptual Losses for Real-Time Style Transfer and Super-Resolution (2016), J. Johnson. [pdf] [code]

Acknowledgement

Thank you for all your contributions.

License

MIT

To the extent possible under law, William Falcon has waived all copyright and related or neighboring rights to this work.