how to find the optimal inputs values of neural network that can give optimal value of predicted output

Dear All;

if i built a neural network model or any other black box models to predict an output value from a set of input value, then i need to find the values of the inputs that can yeild the optimal value of the predcited output , how can we do that ?can we run the stadrard optimization techniques? or shall we use search algorthim such as GA ? How can we do that in MATLAB?

[Bernhard Suhm]

Something isn't quite right with the question "finding values of the inputs that yield the optimal value of the predicted output": for any given input, a model will provide a predicted output, you can't "optimize" the value of the output. What you can optimize are so-called hyperparameters of your model to overall (not for one specific input) provide more accurate predictions. You could use baysian hyperparameter optimization to accomplish that in MATLAB. - Or you can find the set of input values that correspond to a desired "optimal" output value of a model, that sounds more like an optimization problem. - Which one are you interested in?

[Mustafa Al-Nasser]

Thank you for your help

i mean the second one but i did it already thank you again

[Samardeep Sarna]

Hi Mustafa, how did you do it? I also have the same problem.

Answer 1

[Amid Morshedlou]

It is my question too, it would be appreciated if you can help me with it.