Keras GPU 설치

'''Trains a simple convnet on the MNIST dataset.

Gets to 99.25% test accuracy after 12 epochs

(there is still a lot of margin for parameter tuning).

16 seconds per epoch on a GRID K520 GPU.

'''

from __future__ import print_function

import keras

from keras.datasets import mnist

from keras.models import Sequential

from keras.layers import Dense, Dropout, Flatten

from keras.layers import Conv2D, MaxPooling2D

from keras import backend as K

batch_size = 128        # 배치 사이즈 128로 설정.

num_classes = 10        # 정답 라벨링 개수 ( 0 ~ 10 )

epochs = 3             # Epoch 반복 회숫 3회로 설정

# input image dimensions

img_rows, img_cols = 28, 28         # 이미지 가로 세로 배열.

# MNIST 데이터 불러오기

(x_train, y_train), (x_test, y_test) = mnist.load_data()

if K.image_data_format() == 'channels_first':

    x_train = x_train.reshape(x_train.shape[0], 1, img_rows, img_cols)

    x_test = x_test.reshape(x_test.shape[0], 1, img_rows, img_cols)

    input_shape = (1, img_rows, img_cols)

else:

    x_train = x_train.reshape(x_train.shape[0], img_rows, img_cols, 1)

    x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1)

    input_shape = (img_rows, img_cols, 1)

x_train = x_train.astype('float32')

x_test = x_test.astype('float32')

x_train /= 255

x_test /= 255

print('x_train shape:', x_train.shape)

print(x_train.shape[0], 'train samples')

print(x_test.shape[0], 'test samples')

# convert class vectors to binary class matrices

y_train = keras.utils.to_categorical(y_train, num_classes)

y_test = keras.utils.to_categorical(y_test, num_classes)

model = Sequential()

model.add(Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=input_shape))           # Input Layer : 32개의 필터를 가진다.

model.add(Conv2D(64, (3, 3), activation='relu'))                                                # 첫번째 hidden layer이며 64개의 필터를 가진다.

model.add(MaxPooling2D(pool_size=(2, 2)))                                                       # Pooling Layer를 의미한다.

model.add(Dropout(0.25))                                                                        # Dropout을 0.25로 하여 overfitting을 방지한다.

model.add(Flatten())                                                                            # 데이터를 1차원으로 바꿔주는 Layer. 아래의 fully_connected_layer에 연결해주기 위해 1차원으로 바꾸어준다.

model.add(Dense(128, activation='relu'))                                                        # Dense는 Fully Connected Layer를 의미한다.

model.add(Dropout(0.5))                                                                         # Dropout은 0.5로 하여 overfitting을 방지한다.

model.add(Dense(num_classes, activation='softmax'))                                             # Output Layer를 의미하며 activation은 softmax로 분류흘 한다.

model.summary()                                                                                 # 모델의 요약을 보여준다.

model.compile(loss=keras.losses.categorical_crossentropy,

            optimizer=keras.optimizers.Adadelta(),

            metrics=['accuracy'])

model.fit(x_train, y_train,

        batch_size=batch_size,

        epochs=epochs,

        verbose=1,

        validation_data=(x_test, y_test))

score = model.evaluate(x_test, y_test, verbose=0)

print('Test loss:', score[0])

print('Test accuracy:', score[1])