There are many nature inspired optimization algorithms and many of them are maybe arriving very soon. Isn't it the time to STOP introducing new algorithms now? Or maybe there is yet an unseen wise element of the nature deserve to simulate (behaviorally or/and structurally)! If yes, what is the next step?
Thank you for raising this question. it is a very interesting one.
They are many around and some of them behave the same way same others, but have another name. In my opinion, it is a bit of the jungle.
Generative hyper-heuristics appears to be the next step. A metaheuristics generates another metaheuristics for a problem. Read some of my publications.
Many of them are not human-understandable though. As tree-GP can bloat and generate an algorithm that can be very challenging for the human brain to understand. Linear GP, MEP and grammatical evolution as well as graph-based GP has shown some promising results.
Best wishes,
Patricia
Thank you for raising this question. it is a very interesting one.
They are many around and some of them behave the same way same others, but have another name. In my opinion, it is a bit of the jungle.
Generative hyper-heuristics appears to be the next step. A metaheuristics generates another metaheuristics for a problem. Read some of my publications.
Many of them are not human-understandable though. As tree-GP can bloat and generate an algorithm that can be very challenging for the human brain to understand. Linear GP, MEP and grammatical evolution as well as graph-based GP has shown some promising results.
Best wishes,
Patricia
Dear Manizheh Ghaem:
Nature inspired optimization algorithms just reflect the possibility to make evolve solution codes tthat represents solutions of real taskss. The evolution of codes can be carried out by many operators. The imiitación of the natural origin processes by some operators constitutes just a possibility. I am author of the Integration of Variables method, devoted to the topic of making evolve populations of solution codes by means systems of operators. Altough my must important works are written in Spanish Language you can read the two works attached, that could help you to be introduced in the topic.
Many thanks for your kind and informative answers. thanks also for good suggestions.
Dear Manizheh Ghaem,
Let me give you my opinion about this. First thing first, it seems like that right now we have too many nature-based optimization algorithms. The bad thing about this is that the novelty of these new methods is only the source of the inspiration itself, but with only slight contribution to the optimization community. If one wants to propose a new algorithm, they should focus more on more interesting and scientifically sound features, such as adaptivity, data-driven optimization, hybrid of metaheuristics and mathematical based technique, or method that exploit the behavior of the problem being solved (such as surrogate model). In my opinion, we have to be careful if we want to develop a new algorithm since many new nature-based optimization algorithms are prone to this scientific fallacy.
By the way, the following is a nice reference if you want to know what is the fallacy of the novel algorithms and why the metaheuristic community should move to better research direction:
http://onlinelibrary.wiley.com/doi/10.1111/itor.12001/abstract
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators