Wikipedia:Articles for deletion/Imperialist competitive algorithm

Imperialist competitive algorithm

(Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL)

More metaphor-inspired metaheuristic cruft. —Ruud 15:23, 25 July 2016 (UTC)[reply]

Why Imperialist Competitve Algorithm article should be kept in Wikipedia?

I think the removal of this page from WikiPedia is not a good decision. First of all, Imperialist Competitive Algorithm is a widely used and cited algorithm. The first paper that introduced this algorithm has been cited more than 1,000 times. Also in Google Scholar there are more than 3,000 papers that have used or referred to this algorithm. So by just attaching a label "metaphor-inspired metaheuristic cruft" and deleting an article from WikiPedia, I think a decision is beeing made on behalf of a whole community of researchers.

Before making a decision, we should first bring a scientific definition and criteria for calling the work thousands of researchers, "metaphor-inspired metaheuristic cruft". Let's first ask this question: What makes an algorithm to be called "metaphor-inspired metaheuristic cruft"? What is the criteria?

Some suggestions:

Is that the age of and algorithm? So Genetic Algorithm can be called novel because it was proposed in 1950s and another algorithm proposed in 2,000 is not? If yes, then what is the specific year at which we should cut and label all the newer algorithms "metaphor-inspired metaheuristic cruft"? In this case, how should we label "Particle Swarm Optimization" and "Ant Colony"?
Is that the performance of the algorithm of a set of benchmark problems?
Or there is something else?

As we see, calling an algorithm "metaphor-inspired metaheuristic cruft" is not a personal decision we make for ourselves? It is a claim that requires expertise and level of research that leads to coming up with a list of "scientific criteria".

I have been the reviewer of tens of papers in the area of evolutionary computation. So let me share my experience with you. The result is not a set of comprehensive criteria, but using this you can easily identify more than 90% of "metaphor-inspired metaheuristic crufts". The criteria is simple: any algorithm that fits into one the following categories can be labeled as "metaphor-inspired metaheuristic cruft" (the definition does not claim that if an algorithm does not satisfy any of the following conditions, it is definitely novel).

Criterion 1) The source of inspiration is a subgroup of a previously proposed algorithm. For example, after ant colony, another algorithm that uses "American Ant", "Europian ant" and so on, will not be a new algorithm and will be "metaphor-inspired metaheuristic cruft".

Criterion 2) The source of inspiration comes from a scientific fact that does not include "gradual improvement and solution finding (optimization) in nature". For example, Ant Colony has a source of inspiration that comes from the real process that ants apply to find the "optimal" path. However, on the other hand, there are algorithms that model some laws in Pysics. For example "A Gravitational Search Algorithm" which is said to be "based on the Newtonian gravity and the laws of motion" is not using a valid source of inspiration for the algorithm design, even if at the end there is an algorithm that works. The optimization algorithms should be based on a source of inspiration that actually does optimization in nature (E.g. the Evolution is actually optimizing the nature and it is a valid source of inspiration for an algorithm and ant really and really run an optimization process in their daily work and that is the reason they can be a valid source of inspiration for an algorithm - ACO).

Here it is not claimed that the above-mentioned criteria are comprehensive and enable us to easily find and label "metaphor-inspired metaheuristic cruft". However, using just the mentioned two criteria you can find and filter many of the "cruft"s. Then use other methods to work on the few remaining ones. As you see, here instead of labeling this and that, I am talking about criteria.

Not for a few reason, Imperialist Competitive Algorithm is different from many of the proposed algorithm.

Criterion 1) The source of inspiration is a new labeling of a previously proposed source of inspiration. It is based on the theory of social Darwinism which expansion of the Darwins theory to sociology, politics, and history. It is the first and the only major algorithm that is based on a source of inspiration that is not in the category of natural science and still has a strong connection with scientific theories. So the source of inspiration of the algorithm is totally new. The way GA, in the reverse application, is used to simulate artificial life, ICA can be used to model artificial history and social evolution. Giving life to concepts life "country" and "nations" (based on the theories of socio-political evolution), the source of inspiration of ICA is unique and deserves attention and is one of the reasons behind its success.

Criterion 2) Unlike many of the "crufts", ICA is not enforcing a fake connection between the algorithm world and the source of inspiration. It is actually based on a process that is basically doing optimization in nature and human history. As mentioned, ICA is based on Social Darwinism. Extreme Social Darwinists even justify the wars as something useful for optimization human civilization. The source of inspiration for ICA is based on an actual optimization process. What makes it hard for some to understand it is that the source is not categorized in natural science which is the case with many of the similar works. But ICA is simply GA of history and concepts with a totally different point of view that is based on socio-political evolution.

On the other hand, ICA has been trusted and used by thousands of researchers in solving thousands of problems that are published in a few thousand papers. Actually, ICA is among the few algorithms that have a unique real source of inspiration and has been widely used and tested by the researchers.

Any decision to delete this article (or any article) is something that should be made based on a set of criteria. We might agree or disagree with the defined set of criteria. We might also add new conditions. As long as we use criteria, not the names, to make decisions, then the decision is fine. A good criterion should have the potential to be applied to any algorithm, regardless of the name, age, fame and so on. The criteria should be logical and clear.

If we ignore using criteria and just use poetic words and phrases as labels then we are not having a scientific decision, it is indeed a personal decision on behalf of a whole community. The criteria must be so clear and precise that even we apply it to famous algorithms like Genetic Algorithms and Ant Colony Optimization and treat them using the same conditions we have for the others. Because, science is the area of criteria-based reasoning, not making a decision based on examples and without reaching a criterion. If there is any simple, clear and precise definition that classifies the work of about 3,000 researchers as "metaphor-inspired metaheuristic cruft", we should first state it, justify it and then use it to claim that an algorithm is cruft, referencing criterion number X. To separate a good and bad scientific work, the method and reasoning should be scientific itself. We can not help science by approaching it with a non-scientific labeling and without having well-defined criteria.

So if there is any criterion that classifies ICA in "metaphor-inspired metaheuristic cruft" we should have it first, before making any decision.