Hello,

What do you mean by he convergence curve? it looks like some metric for the y-axis since usually the x-axis for a graph is the number of iterations.

I will be looking forward to hearing from you.

@Samer Attrah, I would like to mean that in some representation convergence curve of the evolutionary algorithm is represented using no.of iterations along x axis and fitness value along y axis, but some cases the number of functions evaluations is taken along x axis.

My question is which approach is convenient?

I am sorry I did not notice it is for the evolutionary algorithm, I thought it was for Deep learning, I don't know about this one.

@Samer Attah, thank you.

What if no of iterations are less but computational time per iteration is high, how would you compare?

For comparisons of population based algorithms the number of iterations is not pertinent for different algorithms usually use different population sizes. So it is better to plot the curves (best_result vs number_of_evaluations).

Definitely use the number of evaluations if you can't guarantee that the algorithms consume the same number of evaluations per generation/iteration.

Article Maximum number of generations as a stopping criterion consid...