具有二维阵列的 Keras 网络

Question

我有以下输入数组，我想将其提供给 CNN

import numpy as np 
from keras.layers import Input, Dense, Conv2D
#dummy data for 1d array
width = 5
levels = 4
inputArray =np.random.randint(low=0, high=levels, size=20)
inputArray_ = inputArray.reshape(-1,width) #reshape to 2d array

网络的第一层是：

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

我收到以下错误：

ValueError: Layer conv2d_13 was called with an input that isn't a symbolic tensor.   Received type: <class 'numpy.ndarray'>. Full input: [array([[3, 1, 1, 1, 1],
   [0, 1, 3, 1, 3],
   [0, 0, 2, 0, 0],
   [1, 3, 2, 2, 0],
   [1, 1, 2, 1, 1],
   [0, 3, 1, 3, 0],
   [3, 1, 2, 3, 2],
   [1, 2, 0, 2, 2],
   [3, 2, 2, 1, 0],
   [1, 2, 1, 3, 1],
   [0, 3, 2, 3, 0],
   [0, 3, 1, 0, 0],
   [2, 2, 0, 0, 2],
   [2, 2, 1, 0, 1],
   [3, 3, 0, 3, 1],
   [0, 0, 3, 1, 0],
   [1, 3, 1, 2, 2],
   [1, 0, 3, 2, 2],
   [3, 1, 2, 1, 2],
   [3, 0, 3, 3, 1]])]. All inputs to the layer should be tensors.

Answer 1

您需要将数组转换为张量。

import numpy as np
import tensorflow.keras.backend as K
#dummy data for 1d array
width = 5
levels = 4
inputArray =np.random.randint(low=0, high=levels, size=20)
inputArray_ = inputArray.reshape(-1,width)
print(type(inputArray_))



inputTensor_ = K.constant(inputArray_)

print(type(inputTensor_))
print(inputTensor_)

Answer 2

您应该将Input 层而不是inputArray_ 传递给Conv2D 层。在定义神经网络的架构时，我们只传递占位符而不传递实际值。

inputArray =np.random.randint(low=0, high=levels, size=20)
inputArray_ = inputArray.reshape(-1,width, 1) #Conv2D accepts 3D array

input_data = Input(name='the_input', shape=(None, width, 1), dtype='float32')
x = Conv2D(16, (3, 3), activation='relu', padding='same')(input_data)