Question

我已经在Keras中实现了一个简单的变分自动编码器，在编码器和解码器中有2个卷积层。代码如下所示。现在，我通过两个跳过连接（类似于U-Net）扩展了我的实现。在下面的代码中，跳过连接分别命名为merge1和merge2。没有跳过连接，一切正常，但是使用跳过连接，我收到以下错误消息：

ValueError：图形已断开：无法获得张量的值 Tensor（“ encoder_input：0”，shape =（?, 64，80，1），dtype = float32）在层“ encoder_input”。访问了以下先前的层没问题：[]

我的代码有问题吗？

import keras
from keras import backend as K
from keras.layers import (Dense, Input, Flatten)
from keras.layers import Conv2D, Lambda, MaxPooling2D, UpSampling2D, concatenate
from keras.models import Model
from keras.layers import Reshape
from keras.losses import mse

def sampling(args):
    z_mean, z_log_var = args
    batch = K.shape(z_mean)[0]
    dim = K.int_shape(z_mean)[1]
    epsilon = K.random_normal(shape=(batch, dim))
    return z_mean + K.exp(0.5 * z_log_var) * epsilon

image_size = (64,80,1)
inputs = Input(shape=image_size, name='encoder_input')

conv1 = Conv2D(64, 3, activation='relu', padding='same')(inputs)
pool1 = MaxPooling2D(pool_size=(2, 2))(conv1)
conv2 = Conv2D(128, 3, activation='relu', padding='same')(pool1)
pool2 = MaxPooling2D(pool_size=(2, 2))(conv2)

shape = K.int_shape(pool2)

x = Flatten()(pool2)
x = Dense(16, activation='relu')(x)
z_mean = Dense(6, name='z_mean')(x)
z_log_var = Dense(6, name='z_log_var')(x)

z = Lambda(sampling, output_shape=(6,), name='z')([z_mean, z_log_var])
encoder = Model(inputs, [z_mean, z_log_var, z], name='encoder')

latent_inputs = Input(shape=(6,), name='z_sampling')
x = Dense(16, activation='relu')(latent_inputs)
x = Dense(shape[1] * shape[2] * shape[3], activation='relu')(x)
x = Reshape((shape[1], shape[2], shape[3]))(x)

up1 = UpSampling2D((2, 2))(x)
up1 = Conv2D(128, 2, activation='relu', padding='same')(up1)
merge1 = concatenate([conv2, up1], axis=3)

up2 = UpSampling2D((2, 2))(merge1)
up2 = Conv2D(64, 2, activation='relu', padding='same')(up2)
merge2 = concatenate([conv1, up2], axis=3)

out = Conv2D(1, 1, activation='sigmoid')(merge2)

decoder = Model(latent_inputs, out, name='decoder')

outputs = decoder(encoder(inputs)[2])
vae = Model(inputs, outputs, name='vae')

def vae_loss(x, x_decoded_mean):
    reconstruction_loss = mse(K.flatten(x), K.flatten(x_decoded_mean))
    reconstruction_loss *= image_size[0] * image_size[1]
    kl_loss = 1 + z_log_var - K.square(z_mean) - K.exp(z_log_var)
    kl_loss = K.sum(kl_loss, axis=-1)
    kl_loss *= -0.5
    vae_loss = K.mean(reconstruction_loss + kl_loss)
    return vae_loss

optimizer = keras.optimizers.Adam(lr=0.001, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.000)
vae.compile(loss=vae_loss, optimizer=optimizer)
vae.fit(train_X, train_X,
        epochs=500,
        batch_size=128,
        verbose=1,
        shuffle=True,
        validation_data=(valid_X, valid_X))

Answer 1

您的解码器将conv1和conv2作为输入。不能仅使用Model(latent_inputs, ...)

创建

您需要Model([inputs, latent_inputs], ...)

在自动编码器中使用跳过连接时，图形断开连接错误

1 个答案: