CNN - fashion mnist

Introduction

Classification of fashion mnist data using convolutional neural networks.

You can modify the model's hyperparameters in setting.py.

also to set training data and testing data path in setting.py.

e.g.

• train_data = 'fashionmnist/fashion-mnist_train.csv'
• test_data = 'fashionmnist/fashion-mnist_test.csv'

Environment:

• python 2.7
• tensorflow
• pandas
• keras

Run

test the model in testing data

python main.py -test

train the model in training data

python main.py -train

run the model in default mode('test')

python main.py

Result

After 20 iterations, the final accuracy of the test data with 10k samples was 91.82%.