Does the input data for neural network for classification should be normalized?

3 次查看（过去 30 天）
afef
afef 2017-6-8
编辑： afef 2017-6-9
Hi everyone, i tried to create neural network for classification using nprtool but i still can't reach a good accuracy however i tried changing the training function and the number of hidden layer neurons and the size of the input so when i was looking for a solution i found that i should normalize the input data so i want to know should i do this with matlab. This is an example of my input data:
please i really need help ?

请先登录，再回答此问题。

采纳的回答

Greg Heath
Greg Heath 2017-6-9
MATLAB automatically normalizes input and target data and unnormalizes output results.
However, I typically normalize my regression inputs and targets to have zero mean and unit standard deviation jn order to locate and remove or modify outliers.
Hope this helps.
Thank you for formally accepting my answer
Greg
  2 个评论
naghmeh moradpoor
naghmeh moradpoor 2017-6-9
Hi Greg, I have seen two formulas for normalisation, could I ask which formula you use for normalisation? Regards, Ngh
afef
afef 2017-6-9
编辑：afef 2017-6-9
Hi Greg , Can i know how do you normalize your data so i can try it? and sorry you didn't answer me for the first part of my question how to get a good accuracy for my neural network ? and did the quality of the input data have any influence on the accuracy?
Thanks

请先登录，再进行评论。

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Sequence and Numeric Feature Data Workflows 的更多信息

标签

产品

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by