bias and variance of trained neural network

Question

Mohamad 2013-12-1

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/108205-bias-and-variance-of-trained-neural-network

评论： Muhammad Adil Raja 2018-12-8

采纳的回答： Greg Heath

Hello How can we save the bias and variance of a trained network? thank you in advance

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Greg Heath 2013-12-2

2
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/108205-bias-and-variance-of-trained-neural-network#answer_116949

在 MATLAB Online 中打开

 [x,t] = simplefit_dataset;
 net = fitnet(4);
 rng(0)
 [ net tr y e ] = train(net,x,t);
 biaserr = mean(e(:))
 varerr  = var(e(:))

HTH

Thank you for formally accepting my answer

Greg

2 个评论
显示无隐藏无

SAM 2017-10-13

hi Greg, how can I calculate bias and variance if I have mean square error ? or i must have err=Targets-predictions

Muhammad Adil Raja 2018-12-8

Hi Greg, I really like this definition of variance as it is calculated over the error and not on the model output alone. However, is there a reference to back it up? Normally the variance is defined as the variance of the model output, which rather sounds non-sensible to me.

Moreover, how to compute these values if I am using fitnet for MIMO (multiple input or multiple output) problems or while using ensembles?

Best,

MA

请先登录，再进行评论。