keras总结帖

早有耳闻keras是一个神器，对于我这种需要在一个月内完成竞赛任务的同时，还要上课的人来说，很方便。之前学习了tensorflow，keras安装时也是基于tensorflow作为backen的。主要学习了莫老师的课程，在此总结一下keras的基本使用方法和技巧。

由于服务器上keras版本较低，需要以下修改方式

# 新版keras中有几个地方修改了：
np_utils.to_categorical(y_test, nb_classes=10) ----> nb_classes变成num_classes
model.fit(X_train,y_train,epochs=2,batch_size=32) -----> nb_epoch变成epoch

实现一个全连接层

######################################################
######################################################
# 数据标准化处理[regression将输入变成一列]
X_train = X_train.reshape(X_train.shape[0], -1) / 255.   # normalize
X_test = X_test.reshape(X_test.shape[0], -1) / 255.      # normalize

#CNN将输入变成[batch,channel,width,height]
X_train = X_train.reshape(-1, 1,28, 28)/255.
X_test = X_test.reshape(-1, 1,28, 28)/255.

#RNN将数据看成 [batch,time_step,input_dim]
X_train = X_train.reshape(-1, 28, 28) / 255.      # normalize
X_test = X_test.reshape(-1, 28, 28) / 255.        # normalize

y_train = np_utils.to_categorical(y_train, num_classes=10)
y_test = np_utils.to_categorical(y_test, num_classes=10)

######################################################
######################################################
from keras.models import Sequential
from keras.layers import Dense

######################################################
######################################################
# 模型建立时不考虑batch
model = Sequential() #建立模型
model.add(Dense(output_dim=1, input_dim=1)) #建立全连接层

# 建立多层模型，第一层定义输入格式，后面网络就不需要定义输入格式了，默认用前一层输出
model = Sequential([
    Dense(output_dim=32, input_dim=784),
    Activation('relu'),
    Dense(10), 
    Activation('softmax'),
])

#建立CNN模型
model = Sequential()

# Conv layer 1 output shape (32, 28, 28)
model.add(Convolution2D(
    nb_filter=32,
    nb_row=5,
    nb_col=5,
    border_mode='same',     # Padding method
    dim_ordering='th',      # if use tensorflow, to set the input dimension order to theano ("th") style, but you can change it.
    input_shape=(1, 28, 28) # (channel, height, width)
))
model.add(Activation('relu'))

# Pooling layer 1 (max pooling) output shape (32, 14, 14)
model.add(MaxPooling2D(
    pool_size=(2, 2),
    strides=(2, 2),
    border_mode='same',    # Padding method
))
model.add(Flatten())
model.add(Dense(1024))
model.add(Activation('relu'))

# 建立一个RNN模型
model.add(SimpleRNN(
    # for batch_input_shape, if using tensorflow as the backend, we have to put None for the batch_size. Otherwise, model.evaluate() will get error.
    batch_input_shape=(None, TIME_STEPS, INPUT_SIZE),       # Or: input_dim=INPUT_SIZE, input_length=TIME_STEPS,
    output_dim=CELL_SIZE,
    unroll=True,
))

# build a LSTM RNN
model.add(LSTM(
    batch_input_shape=(BATCH_SIZE, TIME_STEPS, INPUT_SIZE),       # Or: input_dim=INPUT_SIZE, input_length=TIME_STEPS,
    output_dim=CELL_SIZE,
    return_sequences=True,      # True: output at all steps. False: output as last step.
    stateful=True,              # True: the final state of batch1 is feed into the initial state of batch2
))
# add output layer
model.add(TimeDistributed(Dense(OUTPUT_SIZE)))
######################################################
######################################################
#定义优化算法和损失函数
model.compile(loss='mse', optimizer='sgd') 

# Define your optimizer RMSprop
rmsprop = RMSprop(lr=0.001, rho=0.9, epsilon=1e-08, decay=0.0)
# We add metrics to get more results you want to see
model.compile(optimizer=rmsprop,
              loss='categorical_crossentropy',
              metrics=['accuracy'])

# Define your optimizer Adam
adam = Adam(lr=1e-4)
adam = Adam(LR)
# We add metrics to get more results you want to see
model.compile(optimizer=adam,
              loss='categorical_crossentropy',
              metrics=['accuracy'])

######################################################
######################################################
# 训练模型，减少cost [默认优化loss，让其最小]
cost = model.train_on_batch(X_train, Y_train)
#得到训练模型的参数
W, b = model.layers[0].get_weights()

# cost，原来train_on_batch还可以自己指定batch的
X_batch = X_train[BATCH_INDEX: BATCH_INDEX+BATCH_SIZE, :, :]
Y_batch = y_train[BATCH_INDEX: BATCH_INDEX+BATCH_SIZE, :]
cost = model.train_on_batch(X_batch, Y_batch)
cost = model.train_on_batch(X_batch, Y_batch)

# 模型训练 (两轮，batch_size=32)
model.fit(X_train, y_train, nb_epoch=2, batch_size=32)
model.fit(X_train, y_train, nb_epoch=1, batch_size=32)

######################################################
######################################################
# 计算模型在测试集上的误差以及训练结果
cost = model.evaluate(X_test, Y_test, batch_size=40) #在模型建立时就定义了
Y_pred = model.predict(X_test)
pred = model.predict(X_batch, BATCH_SIZE)

# 在模型建立时就定义好了能输出哪些值
# 不知道为什么，如果不加batch_size=10000,有的数据不会被测试到。也就是说，如果你做测试，还是加上batch_size
loss, accuracy = model.evaluate(X_test, y_test) 
loss, accuracy = model.evaluate(X_test, y_test)
cost, accuracy = model.evaluate(X_test, y_test, batch_size=y_test.shape[0], verbose=False)


######################################################
######################################################
# 保存模型以及导入模型
print('test before save: ', model.predict(X_test[0:2]))
model.save('my_model.h5')  
del model  # 删除现有的模型

# load
model = load_model('my_model.h5')
print('test after load: ', model.predict(X_test[0:2]))