Re: ZZ; Neural Networks 的朋友们看过来# EE - 电子工程
o*e
1 楼
My understanding is that you classifers obtained by SVM (or whatever ML
methods) differ if they are trained on different data sets or different SVM
parameters.
Therefore, you can combine GA and SVM (or whatever ML methods) in 2 ways:
1. Fix the data set, use each classifier trained by some parameters as an
instance of GA. Then find the optimal SVM classifer in the sense of the best
paramters on the data set. People often use GA to find better paramters of a
NN classifer. It is not new.
2. Fix th
methods) differ if they are trained on different data sets or different SVM
parameters.
Therefore, you can combine GA and SVM (or whatever ML methods) in 2 ways:
1. Fix the data set, use each classifier trained by some parameters as an
instance of GA. Then find the optimal SVM classifer in the sense of the best
paramters on the data set. People often use GA to find better paramters of a
NN classifer. It is not new.
2. Fix th