Neural Network in Matlab

2015 11 26
1 个回答
回答已采纳
更新时间：2015 12 2
16 次查看（30 天）

1 个投票

I have input and output data to train neural network. How can I predict the following data using Matlab?

请先登录，再回答此问题。

 采纳的回答

Greg Heath
Greg Heath 2015-11-27

2 个投票

For regression or curve-fitting see documentation and an example using
help fitnet
doc fitnet
For classification or pattern recognition see documentation and an example using
help patternnet
doc patternnet
Otherwise provide more thorough information
Hope this helps.
Greg

8 个评论
显示 6更早的评论 隐藏 6更早的评论

Tien Tran
Tien Tran 2015-11-28
编辑：Tien Tran 2015-11-28
thank for your answer Mr.Greg.
However, my question means that I have input data (such as: Pressure, temperature, gravity), output data (z-factor), with 100 input data, I have corresponding 100 output data. I have used ANN in Matlab to train it. So, if I have other 200 input data without output data ,what can I do to know result?
Greg Heath
Greg Heath 2015-11-28
The obvious answer is to run the input through the trained net.
If the net was trained to prevent overfitting or overtraining an overfit net, the only thing I can think of is to add noise to the original data to estimate it's generalization ability.
Good Luck.
Greg
Tien Tran
Tien Tran 2015-11-29
I have understood. Thank for your answer.
However, How to add bias for input without output? Can you give me a example? Wish you the best
Greg Heath
Greg Heath 2015-11-29
I do not understand why you need to add a bias to anything.Please explain what you are doing. I have a feeling that you didn't understand.
Greg
Tien Tran
Tien Tran 2015-11-30
Yes. Indeed, I haven't yet understood about neural network. My school in Vietnam don't teach this knowledge. I read scientific article and find a lot of research with ANN, so I am finding ANN in Matlab. Can you explain detail about ANN, what is main purpose of ANN?. I must have input and corresponding output to train data. However, if I know output why I have to use ANN? Please, explaining for me this stupid question
Greg Heath
Greg Heath 2015-12-1
The net is a mathematical model for obtaining answers when only the input unknown. The caveat is that the properties of the new input should be similar to those of the training and testing data.
Tien Tran
Tien Tran 2015-12-1
Thank Mr.Greg Heath
I have created a neural network for myself. The train performance, R is equal 1, so is it true? Because I have read that ANN just predict approximately
Greg Heath
Greg Heath 2015-12-2
It is not unusual to obtain R=1 for trivial problems.
Try your hand at the MATLAB examples obtained by using the commands
help nndatsets
doc nndatasets
Hope this helps.
Greg

请先登录，再进行评论。

更多回答（0 个）

类别

在 帮助中心 和 File Exchange 中查找有关 Deep Learning Toolbox 的更多信息

标签

Community Treasure Hunt

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

Start Hunting!

Translated by