Keras 2.0版本运行

Keras 2.0版本运行demo出错：

d:\program\python3\lib\site-packages\ipykernel_launcher.py:8: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(32, (3, 3), activation="relu")`

使用Keras时用到了卷积层Convolution2D( )以及Model.fit( )：

x = Convolution2D(8, 3, 3, activation='relu', border_mode='same')(x)
Model.fit(x_train, x_train, nb_epoch=10, batch_size=256, shuffle=True, validation_data=(x_test, x_test))

遇到了如下错误：

UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(1, (3, 3), padding="same", activation="sigmoid")`
decoded = Convolution2D(1, 3, 3, activation='sigmoid', border_mode='same')(x)

在查看了github上的Keras 2.0发行说明后，发现这是从Keras 1到Keras 2发生的变化．

此处涉及到的有：
Convolution* 层被重新命名 Conv*　；
border_mode - > padding　；
nb_epoch - > epochs；
kernel_size可以设置为一个整数，例如Conv2D(10, 3)相当于Conv2D(10, (3, 3))；

因此，

x = Convolution2D(8, 3, 3, activation='relu', border_mode='same')(x)
Model.fit(x_train, x_train, nb_epoch=10, batch_size=256, shuffle=True, validation_data=(x_test, x_test))

改为：

x = Conv2D(8, 3, activation='relu', padding='same')(x)
Model.fit(x_train, x_train, epochs=10, batch_size=256, shuffle=True, validation_data=(x_test, x_test))

修正后版本：

from keras.models import *
from keras.layers import *
import sys

input_tensor = Input((height, width, 3))
x = input_tensor
for i in range(4):
    x = Conv2D(32*2**i, 3, activation='relu')(x)
    x = Conv2D(32*2**i, 3, activation='relu')(x)
    x = MaxPooling2D((2, 2))(x)

x = Flatten()(x)
x = Dropout(0.25)(x)
x = [Dense(n_class, activation='softmax', name='c%d'%(i+1))(x) for i in range(4)]
model = Model(inputs=input_tensor, outputs=x)

model.compile(loss='categorical_crossentropy',
              optimizer='adadelta',
              metrics=['accuracy'])

其他更多变化参见：Keras 2.0发行说明
原文：https://blog.csdn.net/akadiao/article/details/80405766

1、Q：ImportError: No module named visualize_util

A：自从2017年之后，visualize_util 变成vis_utils, 并且plot函数重新命名成plot_model.

down voteaccepted

You will have to either fix the code manually, or downgrade Keras.

By looking at the commit history of that module, you can see that it was renamed on February 28, 2017 from visualize_util to vis_utils. The plot function was also renamed to plot_model.