Continuing last weeks’ search, today I’m looking for areas where PSO is lacking. From what I’ve seen so far, I haven’t heard much said against it, other than the mathematics is not truly understood, there can be tendency to early convergence, and it is necessary to know something about the problem space to calibrate the settings of the algorithm … but that’s it, to date. However, I’ve been given the tip of looking at Stochastic Diffusion Search (SDS), so currently having a little hunt through the literature …
Having gone through a number of parpers, gone to sleep on it all …
