深度学习（Deep Learning） 3.Example

深度学习（Deep Learning） 3.Example

---

1 Introduction

• Keras：easy to learn，作者 Francois Chollet现在google工作，keras已成为Tensorflow的官方API。
• Keras means horn（角）in Greek

---

2 MNIST

• 手写数字识别：deep learning中的hello world

model = Sequential()
model.add(Dense( input_dim = 28*28, output_dim = 500))
model.add(Activation('sigmoid'))
model.add(Dense( output_dim = 500))
model.add(Activation('sigmoid'))
model.add(Dense( output_dim = 10))
model.add(Activation('softmax'))

model.compile(loss = 'categorical crossentropy', optimizer = 'adam', metrics = ['accuracy'])
model.fit(x_train, y_train, batch_size = 100, nb_epoch = 20) #将训练数据分成一些mini-batch，初始化参数-选择1st batch-计算L'-更新参数-选择2nd batch-计算L''-……-所有batch选择过，即一个epoch

score = model.evaluate(x_test, y_test)
print('Total loss on Testing Set:', score[0])
print('Accuracy of Testing Set:', score[1])
# result = model.predict(x_test)

• optimizer中：SGD、RMSprop、Adagrad、Adadelta、Adam、Adamax、Nadam
• SGD中batch_size = 1
• batch_size更大时运算速度更快，因为过程中实现了并行运算（GPU加速）