Dropout Neural Network for Classification
显示 更早的评论
Hello,
I'm studying neural network and i want to know how to do dropout in Matlab for classification. There is something called dropout class, but not much information about that in Matlab. I would like to know if it is possible to apply dropout in neural network in Matlab, and what steps should be done? As far as i know, i have to manipulate the neurons inside hidden layer, however the nntools in Matlab looks like cannot be manipulated. I am trying to create custom neural network, but still i am confused how to create and apply dropout. Unfortunately, i couldn't find any information regarding to this in the internet.
Thank you
2 个评论
Emma Reeves
2017-4-27
I have the same problem. Have you found any answer yet?
Greg Heath
2017-4-27
I have been designing classifiers for more than 25 years and have never heard of the term "drop out class". If you can explain to me what it means, I can probably tell you how to deal with it using the MATLAB NN Toolbox.
Hope this helps.
Greg
回答（1 个）
Greg Heath
2017-4-28
Drop out is explained in the MATLAB website. Input values are randomly set to 0.0 with a specified probability. This is a mitigation for overtraining an overfit net.
I have never used it. So, cannot give any advice except to first try it using MATLAB NN example data obtained from MATLAB help and documentation
help nndatasets
doc nndatasets
If you try it, post results.
Hope this helps.
Thank you for formally accepting my answer
Greg
类别
在 帮助中心 和 File Exchange 中查找有关 Deep Learning Toolbox 的更多信息
2016-11-18
2017-4-28
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
Translated by 
