Keras:在不同模型中使用同一图层(权重) [英] Keras: Use the same layer in different models (share weights)

问题描述

快速解答:

这实际上很容易. 这是代码(对于那些不想阅读所有文本的人):

This is in fact really easy. Here's the code (for those who don't want to read all that text):

inputs=Input((784,))
encode=Dense(10, input_shape=[784])(inputs)
decode=Dense(784, input_shape=[10])

model=Model(input=inputs, output=decode(encode))

inputs_2=Input((10,))
decode_model=Model(input=inputs_2, output=decode(inputs_2))

在此设置中，decode_model将使用与model相同的解码层. 如果您训练model，decode_model也将被训练.

In this setup, the decode_model will use the same decode layer as the model. If you train the model, the decode_model will be trained, too.

实际问题:

我正在尝试在Keras中为MNIST创建一个简单的自动编码器:

I'm trying to create a simple autoencoder for MNIST in Keras:

这是到目前为止的代码:

This is the code so far:

model=Sequential()
encode=Dense(10, input_shape=[784])
decode=Dense(784, input_shape=[10])

model.add(encode)
model.add(decode)


model.compile(loss="mse",
             optimizer="adadelta",
             metrics=["accuracy"])

decode_model=Sequential()
decode_model.add(decode)

我正在训练它以学习身份功能

I'm training it to learn the identity function

model.fit(X_train,X_train,batch_size=50, nb_epoch=10, verbose=1, 
          validation_data=[X_test, X_test])

重建非常有趣:

但是我也想看看集群的表示. 将[1,0 ... 0]传递到解码层的输出是什么? 这应该是MNIST中一类的均值".

But I would also like to look at the representations of cluster. What is the output of passing [1,0...0] to the decoding layer ? This should be the "cluster-mean" of one class in MNIST.

为此，我创建了第二个模型decode_model，该模型重用了解码器层. 但是，如果我尝试使用该模型，它会抱怨:

In order to do that I created a second model decode_model, which reuses the decoder layer. But if I try to use that model, it complains:

异常:检查时出错:预期density_input_5具有形状(None，784)但具有形状(10，10)的数组

Exception: Error when checking : expected dense_input_5 to have shape (None, 784) but got array with shape (10, 10)

这似乎很奇怪.它只是一个密集的层，矩阵甚至无法处理784个昏暗的输入. 我决定看一下模型摘要:

That seemed strange. It's simply a dense layer, the Matrix wouldn't even be able to process 784-dim input. I decided to look at the model summary:

____________________________________________________________________________________________________
Layer (type)                     Output Shape          Param #     Connected to                     
====================================================================================================
dense_14 (Dense)                 (None, 784)           8624        dense_13[0][0]                   
====================================================================================================
Total params: 8624

它已连接到density_13. 跟踪层的名称很困难，但这看起来像编码器层.当然，整个模型的模型摘要是:

It is connected to dense_13. It's difficult to keep track of the names of the layers, but that looks like the encoder layer. Sure enough, the model summary of the whole model is:

____________________________________________________________________________________________________
Layer (type)                     Output Shape          Param #     Connected to                     
====================================================================================================
dense_13 (Dense)                 (None, 10)            7850        dense_input_6[0][0]              
____________________________________________________________________________________________________
dense_14 (Dense)                 (None, 784)           8624        dense_13[0][0]                   
====================================================================================================
Total params: 16474
____________________

显然，这些层是永久连接的. 奇怪的是，我的decode_model中没有输入层.

Apparently the layers are permanently connected. Strangely there is no input layer in my decode_model.

如何在Keras中重用图层? 我看过功能性API，但是那里的层也融合在一起了.

How can I reuse a layer in Keras ? I've looked at the functional API, but there too, layers are fused together.

推荐答案

哦，没关系.

我应该已经阅读了完整的功能性API: https://keras.io/getting-started/functional-api-指南/#shared-layers

I should have read the entire functional API: https://keras.io/getting-started/functional-api-guide/#shared-layers

以下是其中一项预测(可能仍然缺少一些培训):

Here's one of the predictions (maybe still lacking some training):

我猜这可能是3? 至少现在可以了.

I'm guessing this could be a 3 ? Well at least it works now.

对于那些有类似问题的人， 这是更新的代码:

And for those with similar problems, here's the updated code:

inputs=Input((784,))
encode=Dense(10, input_shape=[784])(inputs)
decode=Dense(784, input_shape=[10])

model=Model(input=inputs, output=decode(encode))


model.compile(loss="mse",
             optimizer="adadelta",
             metrics=["accuracy"])

inputs_2=Input((10,))
decode_model=Model(input=inputs_2, output=decode(inputs_2))

我只编译了其中一个模型. 对于训练，您需要编译模型，对于不必要的预测.

I only compiled one of the models. For training you need to compile a model, for prediction that is not necessary.

这篇关于Keras:在不同模型中使用同一图层(权重)的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！