最近几年，随着AlphaGo的崛起，深度学习开始出现在各个领域，比如无人车、图像识别、物体检测、推荐系统、语音识别、聊天问答等等。因此具备深度学习的知识并能应用实践，已经成为很多开发者包括博主本人的下一个目标了。
目前最流行的框架莫过于Tensorflow了，但是只要接触过它的人，就知道它使用起来是多么让人恐惧。Tensorflow对我们来说，仿佛是一门高深的Deep Learning学习语言，需要具备很深的机器学习和深度学习功底，才能玩得转。
Keras正是在这种背景下应运而生的，它是一个对开发者很友好的框架，底层可以基于TensorFlow和Theano，使用起来仿佛是在搭积木。只要不停的添加已有的“层”，就可以实现各种复杂的深度网络模型。
因此，开发者需要熟悉的不过是两点：如何搭建积木？都有什么积木可以用？
安装
安装的步骤直接按照官方文档来就行了，我笔记本的环境已经杂乱不堪，没有办法一步一步记录安装配置了。主要是安装python3.6，然后各种pip install就行了。
参考文档：http://keras-cn.readthedocs.io/en/latest/for_beginners/keras_linux/
基础概念
在使用Keras前，首先要了解Keras里面关于模型如何创建。在上面可爱的小盆友的图片中，想要把积木罗列在一起，需要一个中心的木棍。那么Sequential就可以看做是这个木棍。
剩下的工作就是add不同的层就行了：
model = Sequential()
model.add(Dense(32, input_shape=(784,)))
model.add(Activation('relu'))
建立好model后，相当于我们定义好了逻辑模型。此时就需要编译模型，生成对应的代码：
model.compile(optimizer='rmsprop',
loss='binary_crossentropy',
metrics=['accuracy'])
其中optimizer是参数优化的方法，loss是损失函数的定义，metrics是衡量模型效果的指标。
最后再灌入数据进行训练即可:
model.fit(data, labels, epochs=10, batch_size=32)
完整的例子
代码已经上传到github：https://github.com/xinghalo/keras-examples/blob/master/keras-cn/mnist/mnist_mlp.py
很多人hello world跑不通是因为网络问题，不能下载到对应的数据集。我这里把数据集也上传到对应的目录下了，修改对应的path即可。
from __future__ import print_function
import keras
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Dense, Dropout
from keras.optimizers import RMSprop
batch_size = 128
num_classes = 10
epochs = 3
# the data, split between train and test sets
(x_train, y_train), (x_test, y_test) = mnist.load_data("/Users/xingoo/PycharmProjects/keras-examples/keras-cn/mnist/mnist.npz")
x_train = x_train.reshape(60000, 784)
x_test = x_test.reshape(10000, 784)
x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255
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(Dense(512, activation='relu', input_shape=(784,)))
model.add(Dropout(0.2))
model.add(Dense(512, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(num_classes, activation='softmax'))
model.summary()
model.compile(loss='categorical_crossentropy',
optimizer=RMSprop(),
metrics=['accuracy'])
history = 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])
运行效果
Using TensorFlow backend.
/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/importlib/_bootstrap.py:219: RuntimeWarning: compiletime version 3.5 of module 'tensorflow.python.framework.fast_tensor_util' does not match runtime version 3.6
return f(*args, **kwds)
60000 train samples
10000 test samples
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense_1 (Dense) (None, 512) 401920
_________________________________________________________________
dropout_1 (Dropout) (None, 512) 0
_________________________________________________________________
dense_2 (Dense) (None, 512) 262656
_________________________________________________________________
dropout_2 (Dropout) (None, 512) 0
_________________________________________________________________
dense_3 (Dense) (None, 10) 5130
=================================================================
Total params: 669,706
Trainable params: 669,706
Non-trainable params: 0
_________________________________________________________________
2018-05-25 17:15:22.294036: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA
Train on 60000 samples, validate on 10000 samples
Epoch 1/3
128/60000 [..............................] - ETA: 166s - loss: 2.4237 - acc: 0.0703
640/60000 [..............................] - ETA: 38s - loss: 1.8688 - acc: 0.3812
1152/60000 [..............................] - ETA: 23s - loss: 1.5497 - acc: 0.5087
1664/60000 [..............................] - ETA: 18s - loss: 1.3466 - acc: 0.5655
2176/60000 [>.............................] - ETA: 15s - loss: 1.1902 - acc: 0.6167
2688/60000 [>.............................] - ETA: 13s - loss: 1.0736 - acc: 0.6536
3200/60000 [>.............................] - ETA: 12s - loss: 0.9968 - acc: 0.6778
3712/60000 [>.............................] - ETA: 11s - loss: 0.9323 - acc: 0.7002
4096/60000 [=>............................] - ETA: 11s - loss: 0.8971 - acc: 0.7109
...
51328/60000 [========================>.....] - ETA: 0s - loss: 0.0733 - acc: 0.9775
51840/60000 [========================>.....] - ETA: 0s - loss: 0.0733 - acc: 0.9774
52352/60000 [=========================>....] - ETA: 0s - loss: 0.0735 - acc: 0.9774
52864/60000 [=========================>....] - ETA: 0s - loss: 0.0733 - acc: 0.9775
53376/60000 [=========================>....] - ETA: 0s - loss: 0.0736 - acc: 0.9774
53888/60000 [=========================>....] - ETA: 0s - loss: 0.0734 - acc: 0.9775
54400/60000 [==========================>...] - ETA: 0s - loss: 0.0736 - acc: 0.9774
54912/60000 [==========================>...] - ETA: 0s - loss: 0.0740 - acc: 0.9773
55424/60000 [==========================>...] - ETA: 0s - loss: 0.0744 - acc: 0.9772
55936/60000 [==========================>...] - ETA: 0s - loss: 0.0746 - acc: 0.9771
56448/60000 [===========================>..] - ETA: 0s - loss: 0.0749 - acc: 0.9771
56960/60000 [===========================>..] - ETA: 0s - loss: 0.0751 - acc: 0.9772
57472/60000 [===========================>..] - ETA: 0s - loss: 0.0756 - acc: 0.9772
57984/60000 [===========================>..] - ETA: 0s - loss: 0.0754 - acc: 0.9772
58496/60000 [============================>.] - ETA: 0s - loss: 0.0750 - acc: 0.9773
59008/60000 [============================>.] - ETA: 0s - loss: 0.0750 - acc: 0.9773
59520/60000 [============================>.] - ETA: 0s - loss: 0.0749 - acc: 0.9774
60000/60000 [==============================] - 7s - loss: 0.0749 - acc: 0.9774 - val_loss: 0.0819 - val_acc: 0.9768
Test loss: 0.0819479118524
Test accuracy: 0.9768
参考
- Keras中文官方文档：http://keras-cn.readthedocs.io/en/latest/getting_started/sequential_model/
- Keras github examples：https://github.com/keras-team/keras/blob/master/examples/mnist_mlp.py
- 神经网络(一):概念：https://blog.csdn.net/xierhacker/article/details/51771428
- 神经网络(二):感知机：https://blog.csdn.net/xierhacker/article/details/51816484
- 深度学习笔记二：多层感知机（MLP）与神经网络结构：https://blog.csdn.net/xierhacker/article/details/53282038
- 多层感知机：Multi-Layer Perceptron：https://blog.csdn.net/xholes/article/details/78461164