ValueError:形状(100,784)和(4,6836)未对齐:784(dim 1)!= 4(dim 0)

Posted

技术标签:

【中文标题】ValueError:形状(100,784)和(4,6836)未对齐:784(dim 1)!= 4(dim 0)【英文标题】:ValueError: shapes (100,784) and (4,6836) not aligned: 784 (dim 1) != 4 (dim 0) 【发布时间】:2019-04-25 14:27:06 【问题描述】:

更新:我修正了错误,所以我只需要第二个问题的答案!

我是 Python 的新手,在执行任务时遇到了错误。我查找了这个错误,但没有找到我的答案。

所以,这就是我想要做的。

我想构建一个能够预测值的神经网络。 我用于该类的代码如下

# neural network class definition

类神经网络:

#Step 1: initialise the neural network: number of input layers, hidden layers and output layers
def __init__(self, inputnodes, hiddennodes, outputnodes, learningrate):
    #set number of nodes in each input, hidden, output layer
    self.inodes = inputnodes
    self.hnodes = hiddennodes
    self.onodes = outputnodes

    #link weight matrices, wih and who (weights in hidden en output layers), we are going to create matrices for the multiplication of it to get an output
    #weights inside the arrays (matrices) are w_i_j, where link is from node i to node j in the next layer
    #w11 w21
    #w12 w22 etc
    self.wih = numpy.random.normal(0.0,pow(self.inodes,-0.5),( self.hnodes, self.inodes))
    self.who = numpy.random.normal(0.0,pow(self.hnodes,-0.5),( self.onodes, self.hnodes))

    # setting the learning rate
    self.lr = learningrate

    # activation function is the sigmoid function
    self.activation_function = lambda x: scipy.special.expit(x)

    pass

#Step 2: training the neural network - adjust the weights based on the error of the network
def train(self, inputs_list, targets_list):
    #convert input lists to 2d array (matrice)
    inputs = numpy.array(inputs_list, ndmin=2).T
    targets = numpy.array(targets_list, ndmin=2).T

    #calculate signals into hidden layer
    hidden_inputs = numpy.dot(self.wih, inputs)
    #calculate signals emerging from hidden layer
    hidden_outputs = self.activation_function(hidden_inputs)

    #calculate signals into final output layer
    final_inputs = numpy.dot(self.who, hidden_outputs)
    #calculate signals emerging from final output layer
    final_outputs = self.activation_function(final_inputs)
    # output layer error is the (target-actual)
    output_errors = targets -final_outputs
    #hidden layer error is the output_errors, split by weights, recombined at hidden nodes
    hidden_errors = numpy.dot(self.who.T, output_errors)

    #update the weights for the links between the hidden and output layers
    self.who += self.lr * numpy.dot((output_errors*final_outputs * (1.0-final_outputs)),numpy.transpose(hidden_outputs))

    # update the weights for the links between the input and hidden layers
    self.wih += self.lr*numpy.dot((hidden_errors*hidden_outputs*(1.0-hidden_outputs)),numpy.transpose(inputs))

    pass

#Seap 3: giving an output- thus making the neural network perform a guess
def query(self, inputs_list):
    #convert input lists to 2d array (matrice)
    inputs = numpy.array(inputs_list, ndmin=2).T

    #calculate signals into hidden layer
    hidden_inputs = numpy.dot(self.wih, inputs)
    #calculate signals emerging from hidden layer
    hidden_outputs = self.activation_function(hidden_inputs)

    #calculate signals into final output layer
    final_inputs = numpy.dot(self.who, hidden_outputs)
    #calculate signals emerging from final output layer
    final_outputs = self.activation_function(final_inputs)

    return final_outputs

我显然先导入了必要的东西:

import numpy 
#scipy.special for the sigmoid function expit()
import scipy.special

然后我创建了一个神经网络实例:

#number of input, hidden and output nodes
input_nodes = 784
hidden_nodes = 100
output_nodes = 10

#learning rate is 0.8
learning_rate = 0.8

#create instance of neural network
n = neuralNetwork(input_nodes, hidden_nodes, output_nodes, learning_rate)

在此之后,我读取了带有输入和目标的 excel 文件

import pandas as pd
df = pd.read_excel("Desktop\\PythonTest.xlsx")

文件如下所示:

snapshot of file

h、P、D、o 列是输入,EOQ 列是神经网络应该学习的数字。

所以,我首先这样做了:

xcol=["h","P","D","o"]
ycol=["EOQ"]
x=df[xcol].values
y=df[ycol].values

定义 x 和 y 列。 x 是输入,y 是目标。

我现在想根据这些数据训练神经网络,我使用了这些代码行;

# train the neural network
# go through all records in the training data set 
for record in df:
inputs = x
targets = y
n.train(inputs, targets)
pass

这给了我以下错误:

---------------------------------------------------------------------------
  ValueError                                Traceback (most recent call 
  last)
  <ipython-input-23-48e0e741e8ec> in <module>()
  4     inputs = x
  5     targets = y
   ----> 6     n.train(inputs, targets)
  7     pass

  <ipython-input-13-12c121f6896b> in train(self, inputs_list, targets_list)
 31 
 32         #calculate signals into hidden layer
  ---> 33         hidden_inputs = numpy.dot(self.wih, inputs)
 34         #calculate signals emerging from hidden layer
 35         hidden_outputs = self.activation_function(hidden_inputs)

 ValueError: shapes (100,784) and (4,6836) not aligned: 784 (dim 1) != 4 
 (dim 0)

那么两个问题:

    代码出了什么问题? 我想在文件中添加一个额外的列,其中包含训练后的神经网络的猜测。我该如何做到这一点?

在此先感谢您的任何反馈!

干杯

史蒂文

【问题讨论】:

hidden_inputs = numpy.dot(self.wih, inputs) 的参数没有正确的矩阵乘法形状。为了与矩阵a * b 相乘,a 的第二个维度必须与b 的第一个维度匹配。这不会发生,self.wih 的形状为 (100,784),而 inputs 的形状为 (4,6836),所以你得到了错误 (784 != 4)。 assert self.wih.shape[1] == inputs.shape[0] 谢谢!但是我必须在我的代码中在哪里插入“assert self.wih.shape[1] == inputs.shape[0]”?? 断言只是在调用 np.dot 之前的健全性检查。我不知道导致该错误的问题出在哪里,我只是告诉您错误实际上在说什么! 啊哈!谢谢!我发现了错误。我说有 784 个输入和 10 个输出,但事实并非如此。现在唯一的问题:如何在文件中添加一个带有网络猜测的列? 【参考方案1】:

您已经在使用 pandas,因此您可以简单地获取所有输出,并为 pandas df 创建一个新列。

result = [nn.query(input) for input in df]
df['result'] = result

【讨论】:

以上是关于ValueError:形状(100,784)和(4,6836)未对齐:784(dim 1)!= 4(dim 0)的主要内容,如果未能解决你的问题,请参考以下文章

线性回归器无法预测一组值;错误:ValueError:形状(100,1)和(2,1)未对齐:1(dim 1)!= 2(dim 0)

ValueError:无法为具有形状“(?,4)”的张量“Placeholder_36:0”提供形状(4,)的值

ValueError:logits 和标签必须具有相同的形状 ((None, 4) vs (None, 1))

ValueError:检查目标时出错:预期dense_4的形状为(4,)但得到的数组形状为(1,)

用于 LSTM 的 Keras 多元形状

Tensorflow:ValueError:形状必须排名2但排名3