Discussion of article "Population optimization algorithms: Firefly Algorithm (FA)"

New comment
 

New article Population optimization algorithms: Firefly Algorithm (FA) has been published:

In this article, I will consider the Firefly Algorithm (FA) optimization method. Thanks to the modification, the algorithm has turned from an outsider into a real rating table leader.

The firefly algorithm has three rules based on the flickering characteristics of real fireflies. The rules are as follows:

  1. All fireflies will move towards more attractive and bright counterparts.
  2. The degree of attraction of a firefly is proportional to its brightness, which decreases as the distance from another firefly increases due to the fact that the air absorbs light. Therefore, between any two flickering fireflies, the less bright one will move towards the brighter one. If there is no brighter or more attractive counterpart, a firefly will move randomly.
  3. The brightness or light intensity of the firefly is determined by the value of the objective function of the problem.

Fas


    Author: Andrey Dik

     

    Thank you for publishing your research results!

    I like the results and the evaluation methodology - but is there a way to use this optimization technique within the MT5 EA-Optimizer?

    I am coming from the practical side and would like to know how I can use this new research in order to optimize better and more stable EAs.


    Thank you very much!

     
    Eugen Funk #:

    Thank you for publishing your research results!

    I like the results and the evaluation methodology - but is there a way to use this optimization technique within the MT5 EA-Optimizer?

    I am coming from the practical side and would like to know how I can use this new research in order to optimize better and more stable EAs.


    Thank you very much!

    Thanks for the feedback!
    The usual scenario for using such optimization algorithms in trading is self-optimization in Expert Advisors, utilities, indicators, for training neural networks, in adaptive systems.
     
    Andrey Dik #:
    Thanks for the feedback!
    The usual scenario for using such optimization algorithms in trading is self-optimization in Expert Advisors, utilities, indicators, for training neural networks, in adaptive systems.

    Thank you! Would you mind to point me to some example article, which implements "self-optimization"?

     
    Eugen Funk #:

    Thank you! Would you mind to point me to some example article, which implements "self-optimization"?

    https://www.mql5.com/en/search#!keyword=self-optimization&module=mql5_module_articles

    as far as I can tell, the topic of self-optimization in Expert Advisors for MQL5 is not fully disclosed. perhaps I should try to write an article on this topic using one of the optimization algorithms from my articles.

     
    Andrey Dik #:

    https://www.mql5.com/en/search#!keyword=self-optimization&module=mql5_module_articles

    as far as I can tell, the topic of self-optimization in Expert Advisors for MQL5 is not fully disclosed. perhaps I should try to write an article on this topic using one of the optimization algorithms from my articles.

    Thanks for the hints.

    Hmm, what I was basically expecting is a way to run the optimizer with a different optimization algorithm (right now I always use the "fast genetic based algorithm").

    And this looks rather like a script/programm doing everything on the lower level. Not sure however, if I understood this right.

    Would be great to be able to replace the "fast genetic based algorithm" by some customized class implementing the metric calculation (result: float) and the exploration decisions from N previous runs.

    New comment