使用keras根据层名称来初始化网络
keras根据层名称来初始化网络
defget_model(input_shape1=[75,75,3],input_shape2=[1],weights=None): bn_model=0 trainable=True #kernel_regularizer=regularizers.l2(1e-4) kernel_regularizer=None activation='relu' img_input=Input(shape=input_shape1) angle_input=Input(shape=input_shape2) #Block1 x=Conv2D(64,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block1_conv1')(img_input) x=Conv2D(64,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block1_conv2')(x) x=MaxPooling2D((2,2),strides=(2,2),name='block1_pool')(x) #Block2 x=Conv2D(128,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block2_conv1')(x) x=Conv2D(128,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block2_conv2')(x) x=MaxPooling2D((2,2),strides=(2,2),name='block2_pool')(x) #Block3 x=Conv2D(256,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block3_conv1')(x) x=Conv2D(256,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block3_conv2')(x) x=Conv2D(256,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block3_conv3')(x) x=MaxPooling2D((2,2),strides=(2,2),name='block3_pool')(x) #Block4 x=Conv2D(512,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block4_conv1')(x) x=Conv2D(512,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block4_conv2')(x) x=Conv2D(512,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block4_conv3')(x) x=MaxPooling2D((2,2),strides=(2,2),name='block4_pool')(x) #Block5 x=Conv2D(512,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block5_conv1')(x) x=Conv2D(512,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block5_conv2')(x) x=Conv2D(512,(3,3),activation=activation,padding='same', trainable=trainable,kernel_regularizer=kernel_regularizer, name='block5_conv3')(x) x=MaxPooling2D((2,2),strides=(2,2),name='block5_pool')(x) branch_1=GlobalMaxPooling2D()(x) #branch_1=BatchNormalization(momentum=bn_model)(branch_1) branch_2=GlobalAveragePooling2D()(x) #branch_2=BatchNormalization(momentum=bn_model)(branch_2) branch_3=BatchNormalization(momentum=bn_model)(angle_input) x=(Concatenate()([branch_1,branch_2,branch_3])) x=Dense(1024,activation=activation,kernel_regularizer=kernel_regularizer)(x) #x=Dropout(0.5)(x) x=Dense(1024,activation=activation,kernel_regularizer=kernel_regularizer)(x) x=Dropout(0.6)(x) output=Dense(1,activation='sigmoid')(x) model=Model([img_input,angle_input],output) optimizer=Adam(lr=1e-5,beta_1=0.9,beta_2=0.999,epsilon=1e-8,decay=0.0) model.compile(loss='binary_crossentropy',optimizer=optimizer,metrics=['accuracy']) ifweightsisnotNone: #将by_name设置成True model.load_weights(weights,by_name=True) #layer_weights=h5py.File(weights,'r') #foridxinrange(len(model.layers)): #model.set_weights() print'havepreparedthemodel.' returnmodel
补充知识：keras.layers.Dense()方法
keras.layers.Dense()是定义网络层的基本方法，执行的操作是：output=activation（dot（input，kernel）+bias。
其中activation是激活函数，kernel是权重矩阵，bias是偏向量。如果层输入大于2，在进行初始点积之前会将其展平。
代码如下：
classDense(Layer):
"""Justyourregulardensely-connectedNNlayer.
`Dense`implementstheoperation:
`output=activation(dot(input,kernel)+bias)`
where`activation`istheelement-wiseactivationfunction
passedasthe`activation`argument,`kernel`isaweightsmatrix
createdbythelayer,and`bias`isabiasvectorcreatedbythelayer
(onlyapplicableif`use_bias`is`True`).
Note:iftheinputtothelayerhasarankgreaterthan2,then
itisflattenedpriortotheinitialdotproductwith`kernel`.
#Example
```python
#asfirstlayerinasequentialmodel:
model=Sequential()
model.add(Dense(32,input_shape=(16,)))
#nowthemodelwilltakeasinputarraysofshape(*,16)
#andoutputarraysofshape(*,32)
#afterthefirstlayer,youdon'tneedtospecify
#thesizeoftheinputanymore:
model.add(Dense(32))
```
#Arguments
units:Positiveinteger,dimensionalityoftheoutputspace.
activation:Activationfunctiontouse
(see[activations](../activations.md)).
Ifyoudon'tspecifyanything,noactivationisapplied
(ie."linear"activation:`a(x)=x`).
use_bias:Boolean,whetherthelayerusesabiasvector.
kernel_initializer:Initializerforthe`kernel`weightsmatrix
(see[initializers](../initializers.md)).
bias_initializer:Initializerforthebiasvector
(see[initializers](../initializers.md)).
kernel_regularizer:Regularizerfunctionappliedto
the`kernel`weightsmatrix
(see[regularizer](../regularizers.md)).
bias_regularizer:Regularizerfunctionappliedtothebiasvector
(see[regularizer](../regularizers.md)).
activity_regularizer:Regularizerfunctionappliedto
theoutputofthelayer(its"activation").
(see[regularizer](../regularizers.md)).
kernel_constraint:Constraintfunctionappliedto
the`kernel`weightsmatrix
(see[constraints](../constraints.md)).
bias_constraint:Constraintfunctionappliedtothebiasvector
(see[constraints](../constraints.md)).
#Inputshape
nDtensorwithshape:`(batch_size,...,input_dim)`.
Themostcommonsituationwouldbe
a2Dinputwithshape`(batch_size,input_dim)`.
#Outputshape
nDtensorwithshape:`(batch_size,...,units)`.
Forinstance,fora2Dinputwithshape`(batch_size,input_dim)`,
theoutputwouldhaveshape`(batch_size,units)`.
"""
@interfaces.legacy_dense_support
def__init__(self,units,
activation=None,
use_bias=True,
kernel_initializer='glorot_uniform',
bias_initializer='zeros',
kernel_regularizer=None,
bias_regularizer=None,
activity_regularizer=None,
kernel_constraint=None,
bias_constraint=None,
**kwargs):
if'input_shape'notinkwargsand'input_dim'inkwargs:
kwargs['input_shape']=(kwargs.pop('input_dim'),)
super(Dense,self).__init__(**kwargs)
self.units=units
self.activation=activations.get(activation)
self.use_bias=use_bias
self.kernel_initializer=initializers.get(kernel_initializer)
self.bias_initializer=initializers.get(bias_initializer)
self.kernel_regularizer=regularizers.get(kernel_regularizer)
self.bias_regularizer=regularizers.get(bias_regularizer)
self.activity_regularizer=regularizers.get(activity_regularizer)
self.kernel_constraint=constraints.get(kernel_constraint)
self.bias_constraint=constraints.get(bias_constraint)
self.input_spec=InputSpec(min_ndim=2)
self.supports_masking=True
defbuild(self,input_shape):
assertlen(input_shape)>=2
input_dim=input_shape[-1]
self.kernel=self.add_weight(shape=(input_dim,self.units),
initializer=self.kernel_initializer,
name='kernel',
regularizer=self.kernel_regularizer,
constraint=self.kernel_constraint)
ifself.use_bias:
self.bias=self.add_weight(shape=(self.units,),
initializer=self.bias_initializer,
name='bias',
regularizer=self.bias_regularizer,
constraint=self.bias_constraint)
else:
self.bias=None
self.input_spec=InputSpec(min_ndim=2,axes={-1:input_dim})
self.built=True
defcall(self,inputs):
output=K.dot(inputs,self.kernel)
ifself.use_bias:
output=K.bias_add(output,self.bias)
ifself.activationisnotNone:
output=self.activation(output)
returnoutput
defcompute_output_shape(self,input_shape):
assertinput_shapeandlen(input_shape)>=2
assertinput_shape[-1]
output_shape=list(input_shape)
output_shape[-1]=self.units
returntuple(output_shape)
defget_config(self):
config={
'units':self.units,
'activation':activations.serialize(self.activation),
'use_bias':self.use_bias,
'kernel_initializer':initializers.serialize(self.kernel_initializer),
'bias_initializer':initializers.serialize(self.bias_initializer),
'kernel_regularizer':regularizers.serialize(self.kernel_regularizer),
'bias_regularizer':regularizers.serialize(self.bias_regularizer),
'activity_regularizer':regularizers.serialize(self.activity_regularizer),
'kernel_constraint':constraints.serialize(self.kernel_constraint),
'bias_constraint':constraints.serialize(self.bias_constraint)
}
base_config=super(Dense,self).get_config()
returndict(list(base_config.items())+list(config.items()))
参数说明如下：
units：正整数，输出空间的维数。
activation:激活函数。如果未指定任何内容，则不会应用任何激活函数。即“线性”激活：a（x）=x）。
use_bias：Boolean，该层是否使用偏向量。
kernel_initializer：权重矩阵的初始化方法。
bias_initializer：偏向量的初始化方法。
kernel_regularizer：权重矩阵的正则化方法。
bias_regularizer：偏向量的正则化方法。
activity_regularizer：输出层正则化方法。
kernel_constraint：权重矩阵约束函数。
bias_constraint：偏向量约束函数。
以上这篇使用keras根据层名称来初始化网络就是小编分享给大家的全部内容了，希望能给大家一个参考，也希望大家多多支持毛票票。