查看单个帖子
旧 2019-11-24, 13:41   #1
poster
高级会员
 
注册日期: 2019-11-21
帖子: 3,006
声望力: 66
poster 正向着好的方向发展
默认 What function to use to transform real-vaued numbers to lables in classification task

https://in.mathworks.com/help/deeplearning/gs/classify-patterns-with-a-neural-network.html explains how to apply multi layer perceptron for classification task. But it is unclear how to obtain the binary valued labels -- what function to use in the last layer. The model outputs real-valued numbers, so how to transform it to binary 0 and 1. In my dataset, the target is labelled either as 1 or 0 unlike the diagram given in the Matlab tutorial. So, my output layer contains 1 node. Once the model output is calculated, I can use a simple threshold function where all numbers greater than equal to 0.5 are labelled as 1 and the rest as zero. However, there must be some other functions or other thresholds as well. Can somebody please help in explaining how to obtain the labels? Thank you.





More...
poster 当前离线   回复时引用此帖