search

arcaulfield / MNIST_Classification

A supervised classification model, which predicts the labels of a Modified MNIST dataset. 1st place in Kaggle competition (https://www.kaggle.com/c/modified-mnist/overview)
0 stars 0 forks source link

Make a deeper network #21

Closed vaquierm closed 5 years ago

vaquierm commented 5 years ago

A deep version of CNN

https://www.kaggle.com/adityaecdrid/mnist-with-keras-for-beginners-99457

Residual networks?

https://towardsdatascience.com/hitchhikers-guide-to-residual-networks-resnet-in-keras-385ec01ec8ff https://towardsdatascience.com/understanding-residual-networks-9add4b664b03

vaquierm commented 5 years ago

Deeper CNN implemented in #22 Works much better. We should now try to implement a residual neural net?

The code for the deeper CNN comes from here

vaquierm commented 5 years ago

There is a nice implementation of a RNN here https://gist.github.com/mjdietzx/0cb95922aac14d446a6530f87b3a04ce

It implements the ResNet-50 network from this paper

vaquierm commented 5 years ago

This has a lot of models implemented in Keras https://medium.com/datadriveninvestor/five-powerful-cnn-architectures-b939c9ddd57b