Hi,
I just uploaded on ResearchGate a paper describing strategies for solving classification problems with MEP.
The link to read the paper is here:
These strategies were already present in MEPX software, but I did not had time to explain them better and to complete the numerical experiments. Now I did that.
I currently compared MEP only against Linear Genetic Programming and Artificial Neural Networks (see PROBEN1 dataset).
It took some time to complete because LGP uses large populations (like 5000 individuals) which means a big running time (in some cases several minutes per run)
The paper is not submitted for publication yet. I want to add more comparison with other GP techniques.
best regards,
Mihai