mxnet CrossEntropy 测试

  

   

    

     

    
    

     

      from __future__ import absolute_import
     
    
   

    

     

    
    

     

      from __future__ import division
     
    
   

    

     

    
    

     

      from __future__ import print_function
     
    
   

    

     

    
    

     

      import os
     
    
   

    

     

    
    

     

      # import sys
     
    
   

    

     

    
    

     

      # import math
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

      import mxnet as mx
     
    
   

    

     

    
    

     

      import mxnet.gluon.loss as gloss
     
    
   

    

     

    
    

     

      # 
     
    
   

    

     

    
    

     

      import numpy as np
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

      a = np.zeros((2,3))
     
    
   

    

     

    
    

     

      b = np.zeros((2,1))
     
    
   

    

     

    
    

     

      pred = mx.nd.array(a)
     
    
   

    

     

    
    

     

      label = mx.nd.array(b)
     
    
   

    

     

    
    

     

      loss = gloss.SoftmaxCrossEntropyLoss()
     
    
   

    

     

    
    

     

      print("loss",loss(pred, label).asnumpy())
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

      def _softmax(x):
     
    
   

    

     

    
    

     

       step1 = x.exp()
     
    
   

    

     

    
    

     

       step2 = step1.sum(axis=1,keepdims=True)
     
    
   

    

     

    
    

     
 return step1 / step2
     
    
   

    

     

    
    

     

      ce = mx.metric.CrossEntropy()
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

      ce.update(label, _softmax(pred))
     
    
   

    

     

    
    

     

      loss = ce.get()
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

      print(loss)
     
    
  
 

常见函数笔记

  

   

    

     

    
    

     

      def softmax(o):
     
    
   

    

     

    
    

     
 """softmax函数."""
     
    
   

    

     

    
    

     

       o_exp = o.exp()
     
    
   

    

     

    
    

     
 return o_exp / o_exp.sum(axis=1, keepdims=True)
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     
 
     
    
   

    

     

    
    

     

      def net(x):
     
    
   

    

     

    
    

     
 """网络模型."""
     
    
   

    

     

    
    

     
 return softmax(nd.dot(x.reshape((-1, num_pixe
     
    
  
 

文章来源: blog.csdn.net，作者：网奇，版权归原作者所有，如需转载，请联系作者。

原文链接：blog.csdn.net/jacke121/article/details/116714213