Question

我定义了以下CNN模型。期望一维矢量输入的长度为501。

model = ml.models.Sequential()
model.add(ml.layers.Conv1D(filters=NUMBER_OF_FILTERS, kernel_size=KERNEL_SIZE, activation=ACTIVATION, input_shape=(None, 501)))
model.add(ml.layers.MaxPooling1D(pool_size=POOL_SIZE, padding='valid'))
model.add(ml.layers.Flatten())
model.add(ml.layers.Dense(HIDDEN_SIZE-1, activation=ACTIVATION))

但是这会引发值错误：

ValueError: The last dimension of the inputs to `Dense` should be defined. Found `None`.

我不确定Flatten为什么不创建(None, x)之类的形状，而是创建(None, None)之类的形状。这里似乎是什么问题？

这是模型摘要：

Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv1d (Conv1D)              (None, None, 50)          250550    
_________________________________________________________________
max_pooling1d (MaxPooling1D) (None, None, 50)          0         
_________________________________________________________________
flatten (Flatten)            (None, None)              0         
=================================================================
Total params: 250,550
Trainable params: 250,550
Non-trainable params: 0
_________________________________________________________________

Answer 1

我已经找到解决方案。我没有正确定义Conv1D层的input_shape，应该改为：

model.add(ml.layers.Conv1D(filters=NUMBER_OF_FILTERS, kernel_size=KERNEL_SIZE, activation=ACTIVATION, input_shape=(501, 1)))

Answer 2

Layerers Flatten将图像的格式从二维数组（a，b）转换为一维数组（aXb）.Layer Pooling输出max_pooling1d（MaxPooling1D）（None，None，50）一二维数组（0,0）。因此，图层Flatten：flatten（Flatten）（无，无）

Keras中的1D CNN：从合并要素平整到密集层会引发ValueError

2 个答案: