In the attached article, a new criterion for early stopping is proposed. Stopping points are determined using criteria which measure generalization loss and training progress. Results suggest that these criteria can improve he generalization ability of symbolic regression functions evolved using Grammar-based GP.
Please refer to the following links
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.391.1712&rep=rep1&type=pdf
http://stats.stackexchange.com/questions/32415/how-to-avoid-overfitting-when-using-crossvalidation-within-genetic-algorithms
In the attached article, a new criterion for early stopping is proposed. Stopping points are determined using criteria which measure generalization loss and training progress. Results suggest that these criteria can improve he generalization ability of symbolic regression functions evolved using Grammar-based GP.
to solve this QUESTION need we know what method you select to cost of GA for classification, if you use liner method you can management mean error of population in GA and when all population near to each other stop GA, but you use nonlinear you can use Test and checking data.
- Badges
- Science topic
- Similar topics
- Cell Biology
- Methods