>>> model.compile(optimizer='adam',
>>>     loss='sparse_categorical_crossentropy',
>>>     metrics=['accuracy'])
>>> model.fit(train_images, train_labels, epochs=5)
Train on 60000 samples
Epoch 1/5
60000/60000 [==============================] - 49s 818us/sample - loss: 0.1377 - accuracy: 0.9566
Epoch 2/5
60000/60000 [==============================] - 47s 777us/sample - loss: 0.0449 - accuracy: 0.9858
Epoch 3/5
60000/60000 [==============================] - 40s 670us/sample - loss: 0.0320 - accuracy: 0.9902
Epoch 4/5
60000/60000 [==============================] - 39s 642us/sample - loss: 0.0247 - accuracy: 0.9919
Epoch 5/5
60000/60000 [==============================] - 50s 826us/sample - loss: 0.0186 - accuracy: 0.9938
10000/1 - 3s - loss: 0.0266 - accuracy: 0.9845

>>> test_loss, test_acc = model.evaluate(test_images, test_labels, verbose=2)
>>> print(test_acc)
0.9845   # 정분류율이 98.45%로 개선