Re-implementing some known deep learning algorithms for learning purposes
In this project I am implementing (mostly unoriginal) deep learning algorithms for learning purposes. For the moment, everything is done in Keras & Theano.
The ideas and a good deal of the code are kindly provided by the following great resources:
- course.fast.ai that gives a very nice introduction to CNN in Keras + Theano, and solves Cats vs Dogs using VGG 16 much better than I can do.
- http://adilmoujahid.com/posts/2016/06/introduction-deep-learning-python-caffe/ (see https://github.com/adilmoujahid/deeplearning-cats-dogs-tutorial) that discusses the same problem in Caffe, using Alexnet.
- https://github.com/heuritech/convnets-keras that implements VGG and Alexnet in Keras + Theano