Open dtoertei opened 5 years ago
Successfully reached >93% test set accuracy with this convnet (Conv2Dsimple2) without data augmentation. Used stratified 3-fold train, dropout, batch normalization and max norm.
Code and description is available here: https://github.com/dtoertei/fashion-mnist
Successfully reached >93% test set accuracy with this convnet (Conv2Dsimple2) without data augmentation. Used stratified 3-fold train, dropout, batch normalization and max norm.
Code and description is available here: https://github.com/dtoertei/fashion-mnist