Neural Networks warning?
4 次查看(过去 30 天)
显示 更早的评论
Hi,
When I train any neural network i get the warning below. It still trains usable networks but I'd like to know what the warning means.
% Warning: 'trainRation' is not a legal parameter.
% > In nntest.param>do_test (line 63)
% In nntest.param (line 6)
% In network/subsasgn>setDivideParam (line 1838)
% In network/subsasgn>network_subsasgn (line 460)
% In network/subsasgn (line 14)
% In NN_Training (line 78)
I'm using the function below to train the networks but I don't know why trainRation is causing the warning.
net = fitnet(current_neuron_count, TRAIN_FCN);
net.divideParam.trainRation = 70/100;
net.divideParam.valRation = 15/100;
net.divideParam.testRation = 15/100;
Thanks
0 个评论
采纳的回答
Walter Roberson
2019-3-9
trainRatio, valRatio, testRatio
no final 'n'. Not trainRation, trainRatio
3 个评论
Walter Roberson
2019-3-9
The versions with 'Ration' would have had those commands ignored, leaving you with the default ratios.
更多回答(1 个)
alsharif taha
2020-12-5
when i train this network i get errors
please help me
clc
clear
close all
p=[1:10 10:10:100];
t= (p.^2);
net=newff(p,t,[3], {'logsig' 'purelin'});
net.divideParam.trainRatio=1;
net.divideParam.testRatio=0;
net.divideParam.valRatio=0;
net.divideParam.lr=0.01;
net.divideParam.min_grad=1e-20;
net.divideParam.goal=1e-30;
net.divideParam.epochs=300;
net=train(net,p,t);
plot([1:100] .^2,'x')
hold on
plot(round(net(1:100)),'o')
plot(p,t, '*g')
legend('real target', 'output of net', 'training samples', 'location', 'north west')
the error msgs are:
Warning: 'min_grad' is not a legal parameter.
Warning: 'min_grad' is not a legal parameter.
Warning: 'min_grad' is not a legal parameter.
although i defined the epochs to 300 while training continues to reach 1000 epochs
i do not know why ? pls help me
1 个评论
Walter Roberson
2020-12-5
min_grad is for https://www.mathworks.com/help/deeplearning/ref/traingdx.html not for divideParam
另请参阅
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!