search

JJ / 2021-cec-deep-g-prop

Deep-G-Prop for the new edition of CEC in 2021
GNU General Public License v3.0
1 stars 1 forks source link

Compare G-Prop with Deep-G-Prop re number of evaluations #13

Closed JJ closed 3 years ago

JJ commented 3 years ago

I don't know if that's in the original paper, maybe @pacastillo remembers or at least the parameters used. The results published by @lulivi are essentially for 200 evaluations (20 individuals by 10 generations). It takes a few minutes in my computer, BTW, so we could really boost this up by multiplying it by a lot.

pacastillo commented 3 years ago

I remember that in easy problems, a small number of generations were enough. But to solve more complex problems, a number around 100 generations was neccesary.

pacastillo commented 3 years ago

Indeed, in the IEEE TNN paper, we used 50, 100, 500 1000 generations along 50, 100, 500, 1000 training epochs

JJ commented 3 years ago

How many repetitions?