Neural Network data after training

Question

Arijit 2011-12-10

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/23594-neural-network-data-after-training

Hi, I am completely new in ANN modeling using Matlab. Currently I am trying to make a feedforward network(using newfit) with 15 inputs and a single output. I am using dividerand to randomly divide my data for training, validation and test.

My question is after training the network, if I want to get the target data and network output, what I have to do?

If possible, can anyone post any sample code for ANN other than those explained in help documentation.

Thanks for your help

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 2011-12-11

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/23594-neural-network-data-after-training#answer_31051

You can find a number of code examples using an older version of MATLAB via searching the Newsgroup using

heath newff clear close

Hope this helps.

Greg

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Arijit 2011-12-12

Thanks Greg a lot for your support and suggestion...

Now I have one more doubt. After training I can plot output vs target values. But I want to have this plot (with error bars associated with output values) and target values.

What should be the code to do this?

Thanks

请先登录，再进行评论。

Answer 2

Greg Heath 2011-12-13

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/23594-neural-network-data-after-training#answer_31232

It is not clear what you want. In general, for an I-H-O MLP, each target and output are O-dimensional. Do you want a plot for each dimension of the output?

For each output variable you can obtain a scatter plot of output vs target. The vertical distance from a 45-deg line is the error. If you want to add error bars indicating uncertainty, you will have to use the same value for each point, i.e., the RMSE.

Hope this helps.

Greg