Talk:List of metaphor-based metaheuristics

NPOV discussion

I believe that the user Ruud Koot has made highly biased title and discussion in this article. The entire presentation is biased towards the fact that all new nature-inspired algorithm after PSO, GA, DE, and ACO are metaphor-based. Based on the author’s negative and possibly personal point of view, every single algorithm after those mentioned should be avoided. The title itself implies the author’s personal negative view of the topic. Would not be better to use a title “List of recent metaheuristics” and then provide a discussion on the possibility of the existence of metaphor-based algorithms? I believe that someone cannot simply call all algorithms proposed in this field metaphor-based. I agree with you that some of them are, but this cannot be reflected as a general statement for all algorithms in Wikipedia. I have seen that the user Marco castellani has made a very similar discussion in the page Swarm intelligence. The authors has changed several other pages in a same manner, which should be avoided for sure. This might be due the fact that he is new to academia and possibly make major changes without a deep understanding of the field. I understand that he might be over-enthusiastic as a beginning PhD student, but trying to spam research by his or her favourites professors or authors onto Wikipedia does not seem to be appropriate as per Wikipedia’s rules.

The Section ‘Criticism of the metaphor methodology’ is scientifically inaccurate and misrepresents the views of many cited authors.

The point of view disseminated in this section and on the Wikipedia articles of various metaheuristics (e.g. Bees algorithm, Artificial bee colony algorithm, Harmony search, Cuckoo search, Firefly algorithm) is not fair and impartial.
Often, the statements do not appear to reflect the views of the cited authors. For example: the authors of references [1] and [2] in List of metaphor-based metaheuristics argue that Harmony Search is a sub-class of Evolutionary Strategies, and call for more scientific rigour in the presentation of new methods and peer reviewing. Reference [3] criticises many metaheuristics articles for ignoring the No Free Lunch Theorem (a point I fully support), and calls as well for mathematical rigour. Quoting [3]
“…. this doesn’t mean that nature-inspired algorithms are going to decline – not while arriving at approximate solutions to our complex modern problems is still the best that we can do. Instead the focus is shifting towards improving our understanding of how existing approaches work and improving their scientific value.”
In the text, references [1], [2] and [3] are cited to support the claim that:
“metaheuristics in general have started to attract criticism in the research community for hiding their lack of novelty behind an elaborate metaphor”.
Also, there are dozens of high-impact journals that publish articles on nature-inspired metaheuristics (6000 overall citations only in 2015), the Journal of Heuristics is a relatively minor and uninfluential one (Impact Factor 1.135, Q1 in AI according to Scimago Journal & Country Rank). The prominence given to the editorial opinion of this journal in the section is disproportionate in relation to the scientific impact of the journal.
It is also worth noticing that the debate on what constitutes ‘novelty’ in science is hotly contested and variegate. Whether one algorithm is a subclass of another, or an independent method on its own, is an extremely personal and endlessly debatable opinion. In the same way, fans endlessly argue on the originality or derivative nature of music genres.
The current presentation on Wikipedia of nature-inspired metaheuristics presents a very one-sided view of the topic (see also points raised in Talk:Swarm intelligence).
The entire presentation of the Swarm Intelligence topic on Wikipedia needs extensive fact checking from an expert administrator.
Marco castellani 1965 (talk) 16:13, 28 August 2016 (UTC)

I've examined the first three sources cited in the "Criticism" section. I believe the claims of two of these sources (Weyland, 2010 and Brownlee & Woodward, 2015) have been misrepresented. I edited the claim to better match what Brownlee & Woodward and Swan et al. are saying, and simply removed the reference to Weyland, because he doesn't seem to be attacking metaphor-inspired metaheuristics in general. In consideration of WP:STRUCTURE, I also moved the criticism out of the lede, and placed the criticism section at the end of the article. I did not carefully check the remaining citations in the "Criticism" section, but taking it as granted that the quotations presented are both accurate and representative, I didn't see any need to change anything.

I have also expanded the scope of the whole article, as I could see no good reason to constrain it to metaphor-inspired metaheuristics published after 2001. I think this limit is a best arbitrary and at worst an attempt by the list's compiler to isolate those algorithms he doesn't like for special criticism.

In light of the above, I have removed the POV-check tag inserted by User:Aheedar. If anyone still thinks there are POV problems with the article, they're welcome to re-add the template and attract further knowledgeable editors in good standing to this discussion. —Psychonaut (talk) 14:23, 31 August 2016 (UTC)

I note that variations of the "Criticism" section were posted as boilerplate text to the ledes of several articles. For the same reasons given above for this article, I reverted these mass insertions. However, in cases where the articles contained algorithm-specifc criticism, I placed it in a dedicated section and added a link to the general criticism discussed in this article. If anyone wishes to debate the merits of this approach, I suggest it's best to do it here rather than separately on each article's talk page. —Psychonaut (talk) 19:28, 31 August 2016 (UTC)

This centralizing of the general criticism is an improvement over having copied-and-pasted boilerplate in each individual article. However, I wonder whether a "List" article such as this one is the most appropriate place. I wonder whether Metaheuristic#Nature-inspired metaheuristics might be a more appropriate place, and to link to that from here (as well as from each individual algorithm article). Michaelmalak (talk) 21:03, 31 August 2016 (UTC)

Yes, perhaps Metaheuristic#Nature-inspired metaheuristics would be a better place to collect the criticism. Though I'm not sure that that section should necessarily be linked to from all metaphor-inspired metaheuristics articles, for the simple reason that the general criticism doesn't necessarily apply to any given algorithm. Many of them, like simulated annealing, are well established and respected, and have been thoroughly tested in a variety of applications. And no doubt there are even some very new algorithms where the authors haven't attempted to obfuscate the inner workings and/or make grandiose claims about the algorithm's effectiveness or generalizability. That's why I suggested linking to the general criticism only in those cases where the algorithm has been specifically named. Though if there's some other objective criterion we could use, I'd be open to discussing it. (As I mentioned upthread, I don't think using a time-based cutoff is objective. We can't just say that all metaphor-inspired algorithms developed before 2001 are "good" and all those developed afterwards should be tagged for criticism.) —Psychonaut (talk) 21:32, 31 August 2016 (UTC)

Agree with being selective about which algorithms link to the centralized criticism. Michaelmalak (talk) 21:42, 31 August 2016 (UTC)

@Psychonaut: I have an objection on the title. The title implies that all the algorithms in the page are metaphor-based, which is now true even if we consider time-based cutoff. In the papers cited in the criticism section, there is a mention of a few algorithms in the list here: Harmonic search, Imperialist search, etc. I think that if we change the title to "List of metaheuristics" or "List of recent metaheuristics", That reflects the main content of this page better and seems more neutral. — Preceding unsigned comment added by Aheedar (talk • contribs) 22:11, 31 August 2016 (UTC)

Aheedar, do you agree with the classification at File:Metaheuristics classification.svg? If so, perhaps the best title for this article (i.e., the one which would necessitate adding or removing the least number of algorithms to it) would be List of naturally inspired metaheuristics. —Psychonaut (talk) 19:46, 1 September 2016 (UTC)

Psychonaut, yes I do, that is one of the best classification I have ever seen. The title that you suggested (List of naturally inspired metaheuristics) well represent the majority of the algorithms on this page. Aheedar (talk) 05:34, 2 September 2016 (UTC)

I believe that User:Psychonaut has done an overall great job. A have a few concerns on about mixing Evolutionary algorithm with Swarm intelligence and Swarm behaviour, since the two are quite distinct. Strictly speaking Swarm intelligence entails full decentralisation: agents interact and self-organise. Evolutionary algorithm usually implies some degree of centralisation (selection and replacement routines). Also, I am not sure Simulated annealing should be included in List of metaphor-based metaheuristics, since Simulated annealing is based on the improvement of one solution, whilst the List of metaphor-based metaheuristics are all based on the concept of population-based search. I would say that the common trait of the algorithms grouped in this page is nature-inspired population-based search, and maybe this should be reflected somehow in the title of the article

Also, I would probably summarise Sörensen's and Glover's criticism, and the statements from the Journal of Heuristics. In the first case, whilst the general point made about the inflation of nature-inspired metaheuristics is valid, I believe that Sörensen and Glover have mixed in the same pot some well-established algorithms like Ant colony optimization algorithms, Bacterial foraging, Artificial bee colony algorithm, Bees algorithm, Firefly algorithm, and more questionable ones. Perhaps a summary of the criticism with a reference to the paper, instead of a full quote, would be more neutral. I also don't think the statement "publication of papers on metaphor-based metaheuristics has been limited to second-tier journals and conferences" is correct. In the second case, I would simply mention something like: "some journals have started to accept articles only if they contain innovative basic ideas that demonstrate a contribution to their field", and reference the Journal of Heuristics's policy. In my view, the Journal of Heuristics is a relatively minor journal in AI (Impact Factor 1.135, Q1 in AI according to Scimago Journal & Country Rank), I am not sure its opinion should be given such emphasis. I would add instead a mention of the criticism made in reference [46], something on the lines: "Brownlee and Woodward [46] pointed out how too many authors focused on generating one tool that could solve all problems better than any other. Unfortunately, it has been proven this Holy Grail does not exist (No_free_lunch_in_search_and_optimization). Despite that, the old misconception of looking for the 'mother of all metaheuristics' is still widespread."

Finally, regarding the Bees algorithm, I would like to revert to the old revision of 23:48, 3 July 2016 (https://en.wikipedia.org/w/index.php?title=Bees_algorithm&oldid=728207831), with the list of applications. Is everybody happy? 3 of the 25 examples concern my work. If you think that this would imply a conflict of interest, I am happy to remove these 3 examples.

Thanks again for the improvements and re-organisation of the various articles. Marco castellani 1965 (talk) 02:35, 1 September 2016 (UTC)

I don't think a full revert is appropriate. For one thing, there are intervening edits that make useful changes to the content and formatting. But more importantly, twenty-five references for just a few applications is overkill; I can see why Ruud interpreted this as citation spamming. Can you cut things down to one or two significant and representative references per application? This is supposed to be an encyclopedia article aimed at a general audience, not an exhaustive survey of scholarly literature. —Psychonaut (talk) 19:41, 1 September 2016 (UTC)

Painting everything with the same brush

Biased title and text are added to this page and many other evolutionary algorithm pages by user Ruud Koot. His ideas are good. But not good enough to be scientific. The fact is that he is right that there are many or at least some algorithms that hide their novelty behind an elaborate metaphor. But the way he tries to address this issue is unscientific. He has good intentions indeed. But not with enough knowledge and expertise to help with solving the main problem. The problem is that since separating good and bad work requires years of research and work in the area, he chooses the easiest path. Simply "based on the year of publication", anything after PSO, GA, DE, and ACO is painted as metaphor-inspired. It is more like racism in the real world, where since identifying good and bad people individually is hard and requires meeting people and talking to many and needs reading about ethnic groups, religions, nations, and so on, the easiest thing which leads to less pain for the brain is to be racist and "classify people by some easy factors" like their skin color or their religion. Because it is so easy and even satisfying to classify people or anything (here algorithms) by just one or two factor (here the age of the algorithm). But the problem is that we are not here to make easy decisions that require less depth of knowledge and less thinking. We should make right decisions, even by incorporating hundreds of factors if needed.

There are many bad works out there and I agree with user Ruud Koot. But is this right to label anything after 2000 metaphor-inspired and warn people about using them, without bringing any specific definition to what we mean, other than just the publication year of the paper? There should be a difference between scientific reasoning and religion-and-church-like thinking where every church and religion claims that it is the only path to the god and heaven, and any newer religion (here algorithm) is the deviation from the right path and is just a bad cult. :)

What I see from user Ruud Koot is that he has understood the problem and this is great. But his work with spamming and deleting or adding a warning message to any article published after 2000 is not helping us with solving the main problem. It is like trying to decrease the number of divorces in the society by banning the marriage. It might help, but is not the solution.

I think with the level of eager that he has for helping Wikipedia, he should help us by stating what he really means and how he thinks and what is the definition of bad and good work he is using to delete the pages. None of the papers he refers to are talking about how to separate good and bad work and they say this explicitly that they do not know how to bring a definition and are trying just to raise a question. So why a paper that is not scientific itself to conclude with a list of criteria at the end, suddenly becomes the bible to ban all newer algorithms published after 2,000? If the edits are not done based on some definitions and criteria, as I said, this will be just painting everything with the same brush. It will not be good for the science no matter how good is our intention.

209.203.71.82 (talk) 02:40, 31 August 2016 (UTC)

Strongly concur. Those of us who have been in the field for a while will find the views expressed by the "Journal of Heuristics" and "4OR" all too eerily reminiscent of the restrictive policies that were applied (especially in Europe) to what are now seen as seminal works in evolutionary computation, and blighted the careers of key innovators. That said, we do need to evaluate new proposals critically, there is a lot of garbage out there. But we should also remain aware of the (US) 1st amendment adage, that the best remedy to wrong talk is more talk. Here, we need to feature devastating criticisms such as those applied to harmony search not only in the primary page, but also here in this page (someone looking at this page only, one would probably think harmony search was beyond reproach). But we should not remove the discussions: the best way of informing new participants in these fields that these directions are not worth following is to prominently explain why.

--Urilarim (talk) 08:22, 1 December 2018 (UTC)

External links modified

Hello fellow Wikipedians,

I have just modified 2 external links on List of metaphor-based metaheuristics. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

• Added archive https://web.archive.org/web/20110719215312/http://www.engr.iupui.edu/~shi/Coference/psopap4.html to http://www.engr.iupui.edu/~shi/Coference/psopap4.html
• Added archive https://web.archive.org/web/20110719215312/http://www.engr.iupui.edu/~shi/Coference/psopap4.html to http://www.engr.iupui.edu/~shi/Coference/psopap4.html

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}} (last update: 5 June 2024).

• If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
• If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 05:37, 22 May 2017 (UTC)

Reference contents do not reflect text

What was reference [13], "Adaptive Harmony Search Algorithm for Design Code Optimization of Steel Structures" did not rebutt or even reference the criticism of Harmonic Search by Weyland [12]. In any case, the statement that it provides "ample evidence" is POV. While I did find another paper that claims to rebutt [12], it is from a very low impact journal, and I found the argument lacking. For this reason, I decided to just delete the sentence. — Preceding unsigned comment added by Tcooke (talk • contribs) 08:37, 16 November 2017 (UTC)

Self promotion and arxiv preprints

It would be great if the IPs edit warring at this page would review WP:COI and WP:RS - Wikipedia isn't a place to promote yourself by posting links to your arxiv preprints. Also have a look at WP:POINT - you are not going to get your way by disrupting Wikipedia. Reverting my edits at random is just going to ensure you keep getting blocked and will get this article locked down so IP editors won't be able to change it. MrOllie (talk) 19:07, 20 October 2022 (UTC)