自定义Keras层失败
我想定制Keras层,实现两个模型的输出分配不同的权重,权重可以训练如下自定义Keras层失败,keras,neural-network,keras-layer,Keras,Neural Network,Keras Layer,我想定制Keras层,实现两个模型的输出分配不同的权重,权重可以训练如下 prediction1=model1.output prediction2=model2.output class WeightedSum(Layer): def __init__(self,**kwargs): super(WeightedSum, self).__init__(**kwargs) def build(self, input_shape): self.wei
prediction1=model1.output
prediction2=model2.output
class WeightedSum(Layer):
def __init__(self,**kwargs):
super(WeightedSum, self).__init__(**kwargs)
def build(self, input_shape):
self.weights =K.variable(np.random.random(1))
self.trainable_weights=[self.weights]
def call(self, two_model_outputs):
return self.weights * two_model_outputs[0] + (1 - self.weights) * two_model_outputs[1]
def compute_output_shape(self, input_shape):
return input_shape[0]
final_pred=WeightedSum()([prediction1,prediction2])
但是我在写作中犯了一个错误,不知道怎么做。
Traceback (most recent call last):
File "test-paper3.py", line 182, in <module>
final_pred=WeightedSum()([prediction1,prediction2])
File "/root/anaconda3/lib/python3.7/site-packages/keras/engine/base_layer.py", line 431, in __call__
self.build(unpack_singleton(input_shapes))
File "test-paper3.py", line 162, in build
self.weights =K.variable(np.random.random(1))
AttributeError: can't set attribute
回溯(最近一次呼叫最后一次):
文件“test-paper3.py”,第182行,在
final_pred=WeightedSum()([prediction1,prediction2])
文件“/root/anaconda3/lib/python3.7/site packages/keras/engine/base\u layer.py”,第431行,在调用中__
自我构建(解包单例(输入形状))
文件“test-paper3.py”,第162行,内部版本
自权重=K.变量(np.随机.随机(1))
AttributeError:无法设置属性
也许Keras是在保护自己,不让你使用它认为是某种保留的单词
尝试以标准方式添加权重,并使用另一个变量名:
def build(self, input_shape):
self.kernel = self.add_weight(name='kernel',
shape=(1,),
initializer='uniform',
#I suggest a constraint here, see below
trainable=True)
#this works as an initializer for the weights
K.set_value(self.kernel, np.array([0.5]))
#you can use np.random here, but it seems safer to go with 0.5
#this tells keras that the layer is build in fact
super(WeightedSum, self).build(shapes)
当然,您需要在call
方法中将weights
替换为kernel
无关: 我建议您也使用约束将内核保持在0和1之间
from keras.constraints import MinMaxNorm
........
self.kernel = self.add_weight(name='kernel',
shape=(1,),
initializer='uniform',
constraint = MinMaxNorm(0,1)
trainable=True)
........
也许Keras是在保护自己,不让你使用它认为有保留的词 尝试以标准方式添加权重,并使用另一个变量名:
def build(self, input_shape):
self.kernel = self.add_weight(name='kernel',
shape=(1,),
initializer='uniform',
#I suggest a constraint here, see below
trainable=True)
#this works as an initializer for the weights
K.set_value(self.kernel, np.array([0.5]))
#you can use np.random here, but it seems safer to go with 0.5
#this tells keras that the layer is build in fact
super(WeightedSum, self).build(shapes)
当然,您需要在call
方法中将weights
替换为kernel
无关: 我建议您也使用约束将内核保持在0和1之间
from keras.constraints import MinMaxNorm
........
self.kernel = self.add_weight(name='kernel',
shape=(1,),
initializer='uniform',
constraint = MinMaxNorm(0,1)
trainable=True)
........
那么,怎么了?怎么了?其中?回溯(最后一次调用):文件“test-paper3.py”,第182行,在final_pred=WeightedSum()([prediction1,prediction2])文件/root/anaconda3/lib/python3.7/site packages/keras/engine/base_layer.py中,第431行,在call self.build(unpack_singleton(input_shapes))文件“test-paper3.py”中,第162行,内建self.weights=K.variable(np.random.random(1))AttributeError:无法设置attributeSo,有什么问题吗?怎么了?其中?回溯(最后一次调用):文件“test-paper3.py”,第182行,在final_pred=WeightedSum()([prediction1,prediction2])文件/root/anaconda3/lib/python3.7/site packages/keras/engine/base_layer.py中,第431行,在call self.build(unpack_singleton(input_shapes))文件“test-paper3.py”中,第162行,内部版本self.weights=K.variable(np.random.random(1))AttributeError:无法设置属性