Step by step calculation neural network output

2013 8 2
2 个回答
回答已采纳
8 次查看（30 天）

0 个投票

Hi all!
I don't understand what are the different steps for the calculation of a NN output. I tried to do a step-by-step calculation, but i think i missed something: I tried: output = MyTransferFunc( synaptics * input + biases ); But the result is different of output = myneuralnetwork(input);
My question is: why ??
Thanks all !

请先登录，再回答此问题。

 采纳的回答

Greg Heath
Greg Heath 2013-8-3
编辑：Greg Heath 2013-8-3

0 个投票

It depends on what type of network and default properties you used.
The first expression does not include
1. a hidden layer
2. input and target normalization
3. output un-normalization
Search Answers for my code using the keyword
tsettings
Hope this helps.
Thank you for formally accepting my answer
Greg

更多回答（1 个）

AlfredLeMark
AlfredLeMark 2013-8-6

0 个投票

Thank you for your answer. After checking, there is one hidden layer in my neural network, but i considered that in my calcul. About your second and third point, how can I check if my neural netword normalizes the inputs? (same question with the outputs ^^)

3 个评论
显示 1更早的评论 隐藏 1更早的评论

Greg Heath
Greg Heath 2013-8-7
All net properties can be obtained via
net = net % NO SEMICOLON
AlfredLeMark
AlfredLeMark 2013-8-10
Thank you very much! In the properties, I found something: in the Mynet.inputs, there is a nnnetInput, and in this object there is a table called "range". The Help explains: "This property defines the minimum and maximum values of the input data used to configure the input, or zero if the input is unconfigured."
In my case, the range seems like a Nx2 matrix (N = size of an input) with some values (first column = lowest values, and second column largest values) .
The File Help doesn't specifically explain how the inputs are configured. Do you know how please ?
Thank you for your help.
Greg Heath
Greg Heath 2013-8-12
Look and see that MAPMINMAX is a default property. Use help and doc to see what limits that yields.
Also see what happens when MAPMINMAX is replaced by MAPST and ''.
On the other hand, the output data limits may be affected by the output transfer function. For example, if you keep the MAPMINMAX default for the output but specify logsig for the output transfer function, there is a obvious incompatibility. Not sure how that is resolved.
HTH
Greg

请先登录，再进行评论。

类别

在 帮助中心 和 File Exchange 中查找有关 Define Shallow Neural Network Architectures 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