我有一个自动编码器,我尝试使用lambda在中间层中使用输入层的特定值,并产生一个新的张量并发送到下一层,但是会产生此错误:
回溯(最近通话最近一次):
文件“”,第99行,在 型号=模型(输入= [图像,WTM],输出=解码)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ legacy \ interfaces.py”, 第91行,在包装器中 返回func(* args,** kwargs)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ network.py”, 第93行,初始化 self._init_graph_network(* args,** kwargs)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ network.py”, _init_graph_network中的第231行 self.inputs,self.outputs)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ network.py”, _map_graph_network中的第1366行 tensor_index = tensor_index)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ network.py”, 第1353行,在build_map中 node_index,张量索引)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ network.py”, 第1353行,在build_map中 node_index,张量索引)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ network.py”, 第1353行,在build_map中 node_index,张量索引)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ network.py”, 第1353行,在build_map中 node_index,张量索引)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ network.py”, 第1353行,在build_map中 node_index,张量索引)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ network.py”, 第1353行,在build_map中 node_index,张量索引)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ network.py”, 第1353行,在build_map中 node_index,张量索引)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ network.py”, 第1325行,在build_map中 节点= layer._inbound_nodes [node_index]
AttributeError:“ NoneType”对象没有属性“ _inbound_nodes”
这是我的代码,在添加第一个lambda层之后,它将产生此错误!您能告诉我为什么会发生此错误吗?我感谢您的帮助?我需要的是这样的:
wtm={[0,1,1,0],[0,1,1,0],[0,0,0,0],[0,1,0,0]}
我选择wtm[:,i,j]
并生成形状为(28,28,1)
且值为wtm [:,i,j]的新张量。
wt_random=np.random.randint(2, size=(49999,4,4))
w_expand=wt_random.astype(np.float32)
wv_random=np.random.randint(2, size=(9999,4,4))
wv_expand=wv_random.astype(np.float32)
#w_expand[:,:4,:4]=wt_random
#wv_expand[:,:4,:4]=wv_random
x,y,z=w_expand.shape
w_expand=w_expand.reshape((x,y,z,1))
x,y,z=wv_expand.shape
wv_expand=wv_expand.reshape((x,y,z,1))
#-----------------building w test---------------------------------------------
w_test = np.random.randint(2,size=(1,4,4))
w_test=w_test.astype(np.float32)
#wt_expand=np.zeros((1,28,28),dtype='float32')
#wt_expand[:,0:4,0:4]=w_test
w_test=w_test.reshape((1,4,4,1))
wtm=Input((4,4,1))
image = Input((28, 28, 1))
conv1 = Conv2D(64, (5, 5), activation='relu', padding='same', name='convl1e')(image)
conv2 = Conv2D(64, (5, 5), activation='relu', padding='same', name='convl2e')(conv1)
conv3 = Conv2D(64, (5, 5), activation='relu', padding='same', name='convl3e')(conv2)
BN=BatchNormalization()(conv3)
encoded = Conv2D(1, (5, 5), activation='relu', padding='same',name='encoded_I')(BN)
rep=Kr.layers.Lambda(lambda x:Kr.backend.repeat(x,28))
a=rep(Kr.layers.Lambda(lambda x:x[1,1])(wtm))
add_const = Kr.layers.Lambda(lambda x: x[0] + x[1])
encoded_merged = add_const([encoded,a])
#-----------------------decoder------------------------------------------------
#------------------------------------------------------------------------------
deconv1 = Conv2D(64, (5, 5), activation='elu', padding='same', name='convl1d')(encoded_merged)
deconv2 = Conv2D(64, (5, 5), activation='elu', padding='same', name='convl2d')(deconv1)
deconv3 = Conv2D(64, (5, 5), activation='elu',padding='same', name='convl3d')(deconv2)
deconv4 = Conv2D(64, (5, 5), activation='elu',padding='same', name='convl4d')(deconv3)
BNd=BatchNormalization()(deconv4)
#DrO2=Dropout(0.25,name='DrO2')(BNd)
decoded = Conv2D(1, (5, 5), activation='sigmoid', padding='same', name='decoder_output')(BNd)
#model=Model(inputs=image,outputs=decoded)
model=Model(inputs=[image,wtm],outputs=decoded)
decoded_noise = GaussianNoise(0.5)(decoded)
#----------------------w extraction------------------------------------
convw1 = Conv2D(64, (5,5), activation='relu', name='conl1w')(decoded_noise)#24
convw2 = Conv2D(64, (5,5), activation='relu', name='convl2w')(convw1)#20
#Avw1=AveragePooling2D(pool_size=(2,2))(convw2)
convw3 = Conv2D(64, (5,5), activation='relu' ,name='conl3w')(convw2)#16
convw4 = Conv2D(64, (5,5), activation='relu' ,name='conl4w')(convw3)#12
#Avw2=AveragePooling2D(pool_size=(2,2))(convw4)
convw5 = Conv2D(64, (5,5), activation='relu', name='conl5w')(convw4)#8
convw6 = Conv2D(64, (5,5), activation='relu', name='conl6w')(convw5)#4
convw7 = Conv2D(64, (5,5), activation='relu',padding='same', name='conl7w',dilation_rate=(2,2))(convw6)#4
convw8 = Conv2D(64, (5,5), activation='relu', padding='same',name='conl8w',dilation_rate=(2,2))(convw7)#4
convw9 = Conv2D(64, (5,5), activation='relu',padding='same', name='conl9w',dilation_rate=(2,2))(convw8)#4
convw10 = Conv2D(64, (5,5), activation='relu',padding='same', name='conl10w',dilation_rate=(2,2))(convw9)#4
BNed=BatchNormalization()(convw10)
pred_w = Conv2D(1, (1, 1), activation='sigmoid', padding='same', name='reconstructed_W',dilation_rate=(2,2))(BNed)
w_extraction=Model(inputs=[image,wtm],outputs=[decoded,pred_w])
w_extraction.summary()
(x_train, _), (x_test, _) = mnist.load_data()
x_validation=x_train[1:10000,:,:]
x_train=x_train[10001:60000,:,:]
#
x_train = x_train.astype('float32') / 255.
x_test = x_test.astype('float32') / 255.
x_validation = x_validation.astype('float32') / 255.
x_train = np.reshape(x_train, (len(x_train), 28, 28, 1)) # adapt this if using `channels_first` image data format
x_test = np.reshape(x_test, (len(x_test), 28, 28, 1)) # adapt this if using `channels_first` image data format
x_validation = np.reshape(x_validation, (len(x_validation), 28, 28, 1))
#---------------------compile and train the model------------------------------
opt=SGD(momentum=0.99)
w_extraction.compile(optimizer='adam', loss={'decoder_output':'mse','reconstructed_W':'binary_crossentropy'}, loss_weights={'decoder_output': 0.2, 'reconstructed_W': 1.0},metrics=['mae'])
es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=40)
#rlrp = ReduceLROnPlateau(monitor='val_loss', factor=0.1, patience=20, min_delta=1E-4, verbose=1)
mc = ModelCheckpoint('best_model_5x5F_dp_gn.h5', monitor='val_loss', mode='min', verbose=1, save_best_only=True)
history=w_extraction.fit([x_train,w_expand], [x_train,w_expand],
epochs=1,
batch_size=64,
validation_data=([x_validation,wv_expand], [x_validation,wv_expand]),
callbacks=[TensorBoard(log_dir='E:concatnatenetwork', histogram_freq=0, write_graph=False),es,mc])
当我实现它时,显示此错误:
回溯(最近通话最近一次):
文件“”,第1行,在 encode_merged = add_const([encoded,a])
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ engine \ base_layer.py”, 第457行,致电 输出= self.call(输入,** kwargs)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ keras \ layers \ core.py”, 687行,正在通话中 返回self.function(inputs,** arguments)
文件“”,第1行,在 add_const = Kr.layers.Lambda(lambda x:x [0] + x 1)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ tensorflow \ python \ ops \ math_ops.py”, 第866行,在binary_op_wrapper中 返回func(x,y,name = name)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ tensorflow \ python \ ops \ gen_math_ops.py”, 第301行,添加 “添加”,x = x,y = y,名称=名称)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ tensorflow \ python \ framework \ op_def_library.py”, _apply_op_helper中的第787行 op_def = op_def)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ tensorflow \ python \ util \ deprecation.py”, 第488行,在new_func中 返回func(* args,** kwargs)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ tensorflow \ python \ framework \ ops.py”, 第3274行,在create_op中 op_def = op_def)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ tensorflow \ python \ framework \ ops.py”, 第1792行,在 init 中 control_input_ops)
文件 “ D:\ software \ Anaconda3 \ envs \ py36 \ lib \ site-packages \ tensorflow \ python \ framework \ ops.py”, 第1631行,位于_create_c_op中 引发ValueError(str(e))
ValueError:尺寸必须相等,但对于28和4 输入形状为[?,28,28,1],[4,28,1]的'lambda_9 / add'(op:'Add')。
答案 0 :(得分:0)
在keras中,每个图层都应该是keras的Layer类的实例。在下一行
a=rep(wtm[1,1])
您正在选择张量元素而不使用keras层。该行导致错误。您应该将行更改为以下内容以解决该错误。
a=rep(Kr.layers.Lambda(lambda x:x[1,1])(wtm))
由于wtm的形状为[None,4,4,1](您可以尝试打印形状),所以wtm[1,1]
将从第一维中选择一个元素,然后从所选元素中选择第一个元素。如果要获取[1,1]形状数组,可以做的是索引前三个维度。在这里,您还应该注意批处理大小,因此,您想要的是具有形状[Batchsize,1]的数组。可以按照以下步骤进行。
new_wtm = Kr.layers.Lambda(lambda x:x[:,1,1,:])(wtm)
现在new_wtm.shape将为[None,1],并且调用repeat方法将生成具有shape [None,rep,1]的数组。
rep=Kr.layers.Lambda(lambda x:Kr.backend.repeat(x,28))
a=rep(Kr.layers.Lambda(lambda x:x[:,1,1,:])(wtm))
print(a.shape) # [None, 28, 1]
我希望这可以解决问题
要获取形状为[None,28,28,1]的a
,您需要使用平铺方法。
rep=Kr.layers.Lambda(lambda x:Kr.backend.tile(x,[1, 28, 28, 1]))
a_1 = Kr.layers.Lambda(lambda x: x[:, 1, 1, :])(wtm)
a=rep(Kr.layers.Reshape([1,1,1])(a_1))
测试代码
from keras.layers import Input, Concatenate, Activation,GaussianNoise,Dropout,BatchNormalization,MaxPool2D,AveragePooling2D
from keras.layers import Conv2D, AtrousConv2D
from keras.models import Model
from keras.datasets import mnist
from keras.callbacks import TensorBoard
from keras import backend as K
from keras import layers
import matplotlib.pyplot as plt
import tensorflow as tf
import keras as Kr
from keras.optimizers import SGD,RMSprop,Adam
from keras.callbacks import ReduceLROnPlateau
from keras.callbacks import EarlyStopping
from keras.callbacks import ModelCheckpoint
import numpy as np
import pylab as pl
import matplotlib.cm as cm
import keract
from matplotlib import pyplot
from keras import optimizers
from keras import regularizers
from tensorflow.python.keras.layers import Lambda;
#-----------------building w train---------------------------------------------
#wv_expand=np.zeros((9999,28,28),dtype='float32')
wt_random=np.random.randint(2, size=(49999,4,4))
w_expand=wt_random.astype(np.float32)
wv_random=np.random.randint(2, size=(9999,4,4))
wv_expand=wv_random.astype(np.float32)
#w_expand[:,:4,:4]=wt_random
#wv_expand[:,:4,:4]=wv_random
x,y,z=w_expand.shape
w_expand=w_expand.reshape((x,y,z,1))
x,y,z=wv_expand.shape
wv_expand=wv_expand.reshape((x,y,z,1))
#-----------------building w test---------------------------------------------
w_test = np.random.randint(2,size=(1,4,4))
w_test=w_test.astype(np.float32)
#wt_expand=np.zeros((1,28,28),dtype='float32')
#wt_expand[:,0:4,0:4]=w_test
w_test=w_test.reshape((1,4,4,1))
#wt_expand=wt_expand.reshape((1,28,28,1))
#-----------------------encoder------------------------------------------------
#------------------------------------------------------------------------------
wtm=Input((4,4,1))
image = Input((28, 28, 1))
conv1 = Conv2D(64, (5, 5), activation='relu', padding='same', name='convl1e',dilation_rate=(2,2))(image)
conv2 = Conv2D(64, (5, 5), activation='relu', padding='same', name='convl2e',dilation_rate=(2,2))(conv1)
conv3 = Conv2D(64, (5, 5), activation='relu', padding='same', name='convl3e',dilation_rate=(2,2))(conv2)
#conv3 = Conv2D(8, (3, 3), activation='relu', padding='same', name='convl3e', kernel_initializer='Orthogonal',bias_initializer='glorot_uniform')(conv2)
BN=BatchNormalization()(conv3)
#DrO1=Dropout(0.25,name='Dro1')(BN)
encoded = Conv2D(1, (5, 5), activation='relu', padding='same',name='encoded_I',dilation_rate=(2,2))(BN)
#-----------------------adding w---------------------------------------
#-----------------------decoder------------------------------------------------
#------------------------------------------------------------------------------
rep0=Kr.layers.Lambda(lambda x:Kr.backend.tile(x,[1, 28, 28, 1]),name='aux0')
a_0 = Kr.layers.Lambda(lambda x: x[:, 0, 0, :])(wtm)
a0=rep0(Kr.layers.Reshape([1,1,1])(a_0))
rep1=Kr.layers.Lambda(lambda x:Kr.backend.tile(x,[1, 28, 28, 1]),name='aux1')
a_1 = Kr.layers.Lambda(lambda x: x[:, 0, 1, :])(wtm)
a1=rep1(Kr.layers.Reshape([1,1,1])(a_1))
rep2=Kr.layers.Lambda(lambda x:Kr.backend.tile(x,[1, 28, 28, 1]),name='aux2')
a_2 = Kr.layers.Lambda(lambda x: x[:, 0, 2, :])(wtm)
a2=rep2(Kr.layers.Reshape([1,1,1])(a_2))
rep3=Kr.layers.Lambda(lambda x:Kr.backend.tile(x,[1, 28, 28, 1]),name='aux3')
a_3 = Kr.layers.Lambda(lambda x: x[:, 0, 3, :])(wtm)
a3=rep3(Kr.layers.Reshape([1,1,1])(a_3))
add_const1 = Kr.layers.Lambda(lambda x: x[0] + x[1]+x[2]+x[3]+x[4],name='decoder_output')
encoded_merged = add_const1([encoded,a0,a1,a2,a3])
w=Model(inputs=[image,wtm],outputs=encoded_merged)
w.summary()
#----------------------training the model--------------------------------------
#------------------------------------------------------------------------------
#----------------------Data preparesion----------------------------------------
(x_train, _), (x_test, _) = mnist.load_data()
x_validation=x_train[1:10000,:,:]
x_train=x_train[10001:60000,:,:]
#
x_train = x_train.astype('float32') / 255.
x_test = x_test.astype('float32') / 255.
x_validation = x_validation.astype('float32') / 255.
x_train = np.reshape(x_train, (len(x_train), 28, 28, 1)) # adapt this if using `channels_first` image data format
x_test = np.reshape(x_test, (len(x_test), 28, 28, 1)) # adapt this if using `channels_first` image data format
x_validation = np.reshape(x_validation, (len(x_validation), 28, 28, 1))
#---------------------compile and train the model------------------------------
opt=SGD(lr=0.0001,momentum=0.9)
w.compile(optimizer='adam', loss={'decoder_output':'mse'}, metrics=['mae'])
es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=40)
mc = ModelCheckpoint('los4x4_repw.h5', monitor='val_loss', mode='min', verbose=1, save_best_only=True)
history=w.fit([x_train,w_expand], x_train,
epochs=1,
batch_size=32,
validation_data=([x_validation,wv_expand], x_validation))
w.summary()
layer_name = 'lambda_96'
intermediate_layer_model = Model(inputs=watermark_extraction.input,outputs=watermark_extraction.get_layer(layer_name).output)
intermediate_output = intermediate_layer_model.predict([x_test[8000:8001],w_test])
fig = plt.figure(figsize=(20, 20))
rows = 8
columns = 8
first = intermediate_output
for i in range(1, columns*rows +1):
img = intermediate_output[0,:,:,i-1]
fig.add_subplot(rows, columns, i)
plt.imshow(img, interpolation='nearest',cmap='gray')
plt.axis('off')
plt.show()