Variable selection of neural network based on MIV-
2016-08-23
0 0 0
no vote
Other
Earn points
Dombi and others with the average value (MIV) to reflect changes in the neural network weights matrix cases, MIV is considered in the evaluation of neural network variable relevance one of the best indicators, thereby creating a new way to solve this kind of problem. So exploring this type of evaluation index for use in practical work and looking for new evaluation index is a subject worthy of study.
Neural networks in the previous discussion's input is included in the researcher according to the expertise and experience of preselected. In many applications of neural networks, however, because there is no clear basis neural network contains a variable input is difficult to predetermine, if independent variables was also introduced into neural networks that have no importance, will reduce the accuracy of models, so choose a meaningful argument is often a first step in neural network analysis. Select the neural network input there are a number of ways, the basic idea is that s
Neural networks in the previous discussion's input is included in the researcher according to the expertise and experience of preselected. In many applications of neural networks, however, because there is no clear basis neural network contains a variable input is difficult to predetermine, if independent variables was also introduced into neural networks that have no importance, will reduce the accuracy of models, so choose a meaningful argument is often a first step in neural network analysis. Select the neural network input there are a number of ways, the basic idea is that s
matlab
神经网络
bp
基于
筛选
变量
MIV
Related Source Codes
GMSK Linear Receiver
0
0
no vote
NSGA-II algorithm
0
0
no vote
NSGA-III multi-objective optimization algorithm
0
0
no vote
Compressed sensing example
0
0
no vote
CFAR detector example
0
0
no vote
No comment