Difference between r1.5 and the current
@@ -1,4 +1,5 @@
[[pagelist(^(머신러닝스터디/2016))]]
[머신러닝스터디/2016]
[머신러닝스터디/2016/목차]
== 내용 ==* keras 사용
* mnist
내용 ¶
- keras 사용
- mnist
- keras mnist 예제파일 위치: https://s3.amazonaws.com/img-datasets/mnist.pkl.gz
- 코드 실행하면 자동으로 받아짐
- 코드 실행하면 자동으로 받아짐
코드 ¶
from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation
from keras.datasets import mnist
from keras.layers.core import Reshape
import numpy as np
(X_train, y_train), (X_test, y_test) = mnist.load_data()
model = Sequential()
model.add(Reshape((28*28,), input_shape=(28,28)))
model.add(Dense(60000, input_dim=28*28, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(64, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(10, activation='softmax'))
model.compile(loss='categorical_crossentropy',
optimizer='adagrad',
metrics=['accuracy'])
# y_train and y_test is simple integer of 0 to 9
# Need to be transformed to array
y_train_array = np.zeros((60000, 10))
y_test_array = np.zeros((10000, 10))
for i in range(60000):
y_train_array[i][y_train[i]] = 1
for i in range(10000):
y_test_array[i][y_test[i]] = 1
model.fit(X_train, y_train_array,
nb_epoch=3,
batch_size=16)
score = model.evaluate(X_test, y_test_array, batch_size=10000)
# TODO
print(score)
후기 ¶
- 서지혜: 맥에어에서 돌렸더니 엄청 오래걸렸다.. 클라우드 세팅해야될거같음. 아래는 결과물. 다 돌리고 출력을 어떻게 해야할지 모르겠네.
Using Theano backend. Epoch 1/3 60000/60000 [==============================] - 3122s - loss: 14.4306 - acc: 0.1047 Epoch 2/3 60000/60000 [==============================] - 3055s - loss: 14.4370 - acc: 0.1043 Epoch 3/3 60000/60000 [==============================] - 3135s - loss: 14.4453 - acc: 0.1038 10000/10000 [==============================] - 398s [14.461154937744141, 0.10279999673366547]
다음 시간에는 ¶
- Week 6 보기
