Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- from keras.datasets import mnist
- from keras.models import Sequential
- from keras.layers.core import Dense, Activation
- from keras.utils import np_utils
- (X_train, Y_train), (X_test, Y_test), = mnist.load_data()
- X_train = X_train.reshape(60000, 784)
- X_test = X_test.reshape(10000, 784)
- classes = 10
- Y_train = np_utils.to_categorical(Y_train, classes)
- Y_test = np_utils.to_categorical(Y_test, classes)
- input_size = 784
- batch_size = 100
- hidden_neurons = 100
- epochs = 15
- model = Sequential()
- model.add(Dense(hidden_neurons, input_dim=input_size))
- model.add(Activation('sigmoid'))
- model.add(Dense(classes, input_dim=hidden_neurons))
- model.add(Activation('softmax'))
- model.compile(loss='categorical_crossentropy', metrics=['accuracy'], optimizer='sgd')
- model.fit(X_train, Y_train, batch_size=batch_size, nb_epoch=epochs, verbose=1)
- score = model.evaluate(X_test, Y_test, verbose=1)
- print('Dokladnosc testu:', score)
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement