Equation and neural networks
9 views (last 30 days)
Show older comments
i am using neural network to predict the output from 2 inputs, can matlab develop an equation based on the calculation for the predicted output?
Let say, here are my input and output
- input >> output
- 1 1 >> 2
- 2 2 >> 4
and next sequence will be predicted by neural network which is
- 3 3 >> ?
Is there any function that i can use to know the equation used by matlab to calculate next sequence (eg. output= input1 +input2)?
0 Comments
Accepted Answer
Greg Heath
on 30 May 2013
HiddenVariableMatrix = tanh( InputWeightMatrix*InputVariableMatrix + Biasvector1);
OutputVariableMatrix = Biasvector2 + LayerWeightMatrix*HiddenVariableMatrix;
2 Comments
Greg Heath
on 30 May 2013
All calculations are obtained via MATLAB functions which perform all of the calculations within.
See the documentation and demos in the Neural Network Toolbox.
More Answers (2)
KK
on 10 Dec 2016
Hi Greg
I am new to Neural Network and try to learn the theory/formula part such as formulas. Where can I find the equation you pointed out OutputVariableMatrix = Biasvector2 + LayerWeightMatrix*tanh( InputWeightMatrix*InputVariableMatrix + Biasvector1);
under Matlab Neural Netowrk Toolbox?
I try to know other formulas for non-default settings. Could you also recommend some fundamental articles for learning neural network. It is so easy to find the formula about SVM. But I googled a while and could not find a simple clear article about the output equations under neural network. Thanks!
KK
1 Comment
Greg Heath
on 10 Dec 2016
Edited: Greg Heath
on 10 Dec 2016
I don't know where MATLAB has explicitly posted that equation. Replace tanh with tansig and try searching in the NEWSGROUP and ANSWERS.
Try
IW LW TANSIG
Hope this helps.
Greg
Abdelmalek Bellal
on 16 Jan 2022
Hello Greg,
I used a code for the exraction of wieghts and Baises but the collected vector of predictions by runing the ANN model is diffrent from what I am getting by using the mathematical equation. I believe the extracted weights are not the same weights that are use by the model or the equation is incorrcet? I tried multiple times but I am getting abnormal values that are negative and widely far from what I am getting from collceted vector of prediction with runing the model.
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/863940/image.jpeg)
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/863945/image.jpeg)
I added exponential to the output y because I am using Log transform.
Below is the obtained predictions from runing the model and the actual outputs: (reasonable)
![](https://www.mathworks.com/matlabcentral/answers/uploaded_files/863950/image.jpeg)
0 Comments
See Also
Categories
Find more on Sequence and Numeric Feature Data Workflows in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!