简单的自动编码器在许多时期后没有收敛

Question

我正在为 CIFA10 数据集开发自动编码器，但不会在输入端添加噪声（这是第二个目标）。

基于 Convnet 的自动编码器未收敛：任何建议

input_img=Input(shape=(32,32,3))



x=Conv2D(16,(3,3),padding='same',activation='relu')(input_img)

x=MaxPooling2D((2,2),padding='same')(x)

x=Conv2D(8,(3,3),padding='same',activation='relu')(x)

x=MaxPooling2D((2,2),padding='same')(x)

x=Conv2D(8,(3,3),padding='same',activation='relu')(x)

encoded=MaxPooling2D((2,2),padding='same')(x)
x=Conv2D(8,(3,3),padding='same',activation='relu')(encoded)

x=UpSampling2D((2,2))(x)

x=Conv2D(8,(3,3),padding='same',activation='relu')(x)

x=UpSampling2D((2,2))(x)
x=Conv2D(16,(3,3),padding='same',activation='relu')(x) 
x=UpSampling2D((2,2))(x)
decoded=Conv2D(3,(3,3),padding='same',activation='sigmoid')(x)
autoencoder=Model(input_img,decoded)
(x_train,_),(x_test,_)=cifar10.load_data()

x_train=x_train.astype('float32')/255

x_test=x_test.astype('float32')/255

x_train=x_train.reshape(len(x_train),32,32,3)

x_test=x_test.reshape(len(x_test),32,32,3)



autoencoder.compile(optimizer='Adam',loss='binary_crossentropy')
autoencoder.fit(x_train, x_train,

            epochs=50,
            batch_size=64,
            shuffle=True,
            validation_data=(x_test, x_test))

即使在大的时期，我也会遇到损失的高原

我尝试将模型 64-32-16-8:8-16-32-64 与 CIFAR-10 的灰度图像一起使用，但我仍然遇到相同的收敛问题，网络提供模糊输出以及显示 @987654321 @

Answer 1

不确定你的意思。如果我是你，我会尝试更深入的模型以获得更好的性能。例如下面提出的模型（甚至更深的模型）

input_img=Input(shape=(32,32,3))

x=Conv2D(8,(3,3),padding='same',activation='relu')(input_img)
x=Conv2D(8,(3,3),padding='same',activation='relu')(x)

x=MaxPooling2D((2,2),padding='same')(x)

x=Conv2D(16,(3,3),padding='same',activation='relu')(x)
x=Conv2D(16,(3,3),padding='same',activation='relu')(x)

x=MaxPooling2D((2,2),padding='same')(x)

x=Conv2D(32,(3,3),padding='same',activation='relu')(x)
x=Conv2D(32,(3,3),padding='same',activation='relu')(x)

encoded=MaxPooling2D((2,2),padding='same')(x)
x=Conv2D(32,(3,3),padding='same',activation='relu')(encoded)
x=Conv2D(32,(3,3),padding='same',activation='relu')(x)

x=UpSampling2D((2,2))(x)

x=Conv2D(16,(3,3),padding='same',activation='relu')(x)
x=Conv2D(16,(3,3),padding='same',activation='relu')(x)

x=UpSampling2D((2,2))(x)
x=Conv2D(8,(3,3),padding='same',activation='relu')(x) 
x=Conv2D(8,(3,3),padding='same',activation='relu')(x) 
x=UpSampling2D((2,2))(x)
decoded=Conv2D(3,(3,3),padding='same',activation='sigmoid')(x)
autoencoder=Model(input_img,decoded)