sanjay
Full Member level 1
Hi all,
Just a very basic question...
I am training my designed network using, TRAINLM method (levenberg marquadt method), but as it allows one value to adjust (mu)...
Would we still require the value of the learning rate and the momentum..the reason why I ask this is because, when I look at
net.inputWeights{1}.learnParam
I get two values shown namely the learning rate and the momentum, like this
net.inputWeights{1}.learnParam
ans =
lr: 0.0100
mc: 0.9000
So, do these values play a part... seeing the equation for levenberg marquadt.. requires only MU which one can adjust using net.trainParam
Suggestions, thoughts, feedbacks would be appreciated
Regards
Just a very basic question...
I am training my designed network using, TRAINLM method (levenberg marquadt method), but as it allows one value to adjust (mu)...
Would we still require the value of the learning rate and the momentum..the reason why I ask this is because, when I look at
net.inputWeights{1}.learnParam
I get two values shown namely the learning rate and the momentum, like this
net.inputWeights{1}.learnParam
ans =
lr: 0.0100
mc: 0.9000
So, do these values play a part... seeing the equation for levenberg marquadt.. requires only MU which one can adjust using net.trainParam
Suggestions, thoughts, feedbacks would be appreciated
Regards