Handwritten Recognition using Convolutional Neual Network

Training a convolutional neural network for handwritten digits on the famous MNIST dataset.

Dependencies

This code is written in python. To use it you will need:

• python 2.7
• Keras 2.2.2
• numpy 1.15.1
• Tensorflow or Theano

Loading Data

from mnist_loader import load_dataset
train_x, train_y, test_x, test_y = load_dataset()

Training

from ConvolutionalNN import CNN
cnn = CNN()
cnn.train(train_x, train_y, epochs=10, batch_size=32)

Evaluation

score = cnn.evaluate(test_x, test_y)
print('Loss is : ', score[0])
print('Accuracy is : ', score[1])    

Classify a new instance

new_class = cnn.classify(np.array([test_x[0]]))

Save and Load the trained model

cnn.save_model('saved_model')
cnn.load_model('saved_model')