after knowing the optimum weights, how to fix these input weight and bias as constant for remaining training?

Question

prabakaran jayaraman 2015-7-6

1
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/228602-after-knowing-the-optimum-weights-how-to-fix-these-input-weight-and-bias-as-constant-for-remaining

评论： prabakaran jayaraman 2015-7-9

i am using the RNG(0)(thanks to greg) before net command to fix the weight.but it may not be optimal weights.. so i want to train the net and after knowing the optimal weights i want to freeze it. how to fix these input weight and bias as constant for remaining training?

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Greg Heath 2015-7-8

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/228602-after-knowing-the-optimum-weights-how-to-fix-these-input-weight-and-bias-as-constant-for-remaining#answer_185336

If you haven't finished training how can you possibly know that any weights are optimal?

I don't know of a good reason for "fixing" weights. Just set the RNG and let train do it's job.

If you don't like the answer, train again without fixing the RNG. It will automatically be in another state which you can record using s(i) = rng;

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

prabakaran jayaraman 2015-7-9

TEST.m

1. i have finished traing without RNG(0) command. and i got the result. when i use s(i)=rng, it is not recognized. where should i use it. ihave enclosed my program

请先登录，再进行评论。