/B-CNN

Sample code of B-CNN paper (https://arxiv.org/abs/1709.09890) written in Python3+.

Primary LanguagePythonMIT LicenseMIT

---------------------------------------------------------------
Author: Xinqi ZHU
---------------------------------------------------------------
This directory contains sample code of B-CNN paper 
(https://arxiv.org/abs/1709.09890) written in Python3+.

Please ensure you have installed TensorFlow and Keras.

Please make directories metioned in each program
before running them. These directories are used to store 
trained weights and TensorBoard logs.
---------------------------------------------------------------
Files:

CIFAR_10_keras_medium.py: model Base B in the paper 
running on CIFAR-10 dataset

CIFAR_10_keras_medium_dynamic.py: model B-CNN B in 
the paper running on CIFAR-10 dataset

CIFAR_100_keras_vgg16.py: model Base C in 
the paper running on CIFAR-100 dataset

CIFAR_100_keras_vgg16_hierarchy_dynamic.py: model B-CNN C in 
the paper running on CIFAR-100 dataset
---------------------------------------------------------------
To run:

Just type:
python CIFAR_10_keras_medium.py

etc.
---------------------------------------------------------------