How to include a DNN in a Genetic Algorithm searched function

2 次查看（过去 30 天）
Cheng Qiu
Cheng Qiu 2021-11-26
Hello MathWorks Team, I want to use GA to search best input of a certain function so that the function’s output could meet the desired value.
So I create a function to calculate the loss between actual output and desired value. Then I called GA to evaluate the function.Below is my loss function with a dummy function.
function Loss = ForwardPredictionLoss(SPratio,nref)
AA = reshape(nref,4,7);
A = AA*ones(7,4); % ones(7,4) is just some dummy function for test
B = SPratio;
err = (A-B).^2;
Loss = sqrt(sum(sum(err)));
end
and my main script using GA for function input searching:
SPratio = diag(ones(1,4));
nref = ones(28,1);
lb = -ones(28,1);
ub = -lb;
Options = optimoptions(@ga,'PopulationSize',25,'MaxGenerations',20);
[x,fval] = ga(@(nref)ForwardPredictionLoss1(SPratio,nref,FWNnet),28,[],[],[],[],lb,ub,[],Options);
So far, everything goes well. However when i tried to replaced the dummy function with a fixed DNN. I always get some error. Here is my code:
function Loss = ForwardPredictionLoss1(SPratio,nref,FWNnet)
nref1 = nref;
A1 = predict(FWNnet,nref1);
A2 = predict(FWNnet,nref1);
A3 = predict(FWNnet,nref1);
A4 = predict(FWNnet,nref1);
A = [A1 A2 A3 A4];
B = SPratio;
err = (A-B).^2;
Loss = sqrt(sum(sum(err)));
end
clc;clear; close all
load FWN
FWNnet = net;
SPratio = diag(ones(1,4));
nref = ones(32,1);
lb = -ones(32,1);
ub = -lb;
Options = optimoptions(@ga,'PopulationSize',25,'MaxGenerations',20);
[x,fval] = ga(@(nref)ForwardPredictionLoss1(SPratio,nref,FWNnet),32,[],[],[],[],lb,ub,[],Options);
Could you please share the right code for this.
Thanks so much.

请先登录，再回答此问题。

回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Genetic Algorithm 的更多信息

标签

产品


版本

R2020a

Community Treasure Hunt

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

Start Hunting!

Translated by