Talk:Swarm intelligence/Archive 1

Prey
Has anyone read Michael Crichton's Novel Prey? There are some interesting propositions regarding the way that a swarm might learn complex behavior based on basic preditor-prey heuristics. The biggest problem I see with swarm intelligence learning complex behavior, however, is that the smaller the agent is, the less capability it has to internalize and recognize the environment. Dennett refers to creatures that can do this Popperian. What evidence is there that swarms perform Popperian functions, and what is the possibility that this might be possible?

Interesting
The universe seems to be arranged in hierarchies. There are various levels at which you can understand it. The level we are most famililar with is the level of everyday life. Going down we run into the levels of organs, cells, molecular biology, chemistry. Each level has its own laws which work in certian "special cases" with all violations at the "extreams". Complex Adaptive Swarms exhibit similar layered behavior also, at each layer the swarm is made of smaller complex adaptive systems. This layered behavior doesn't appear in swarms made of simple systems. This all seems to imply that either there is no bottom to the layers of the universe(and no TOE) or, the bottom layer is made of smart, adaptive particles.--SurrealWarrior 18:15, 12 July 2005 (UTC)

Why are the ants gone?
I liked that picture. Was it somehow not appropriate for the article?&mdash;GraemeMcRaetalk 19:17, 3 November 2005 (UTC)

references in popular culture
Although they might be related to SI, I would find more appropriate to move this list to the AI article

researchers list
The list is not exhaustive and it contains no useful information. People interested in SI may find useful links to research groups. People interested in the work of a person will first look at her/his personal page, not this article.


 * Agreed. There's a reason why this is the only article I've seen where such a list exists -- it isn't particularly helpful. Perhaps it might justify it's existence if it wasn't primarily composed of red links... --Xanthine 16:03, 1 August 2006 (UTC)


 * I agree too


 * I'm not sure then that we should keep any name. What was the criteria to remove some and leave others. There are some names in the list that are clearly not as important to the field as others that were there. I say the whole section needs to go. If I don't hear from anyone against I'll remove this list soon.

I might want to add, that the Micheal O'Neill that is linked to is not the Michael O'Neill who is doing SI.

Zerg
I'm removing the zerg reference in the article, since the zerg do not fall into "swarm intelligence" category. As stated in the article, in game reality they are centrally controlled (overmind/cerebrate) as they are in real life (game interface). 201.213.16.47 18:41, 28 February 2007 (UTC)

COI and possible reference spam
These were added without any corresponding change to content by User:Ajithabraham Is there any reason to include them? --Ronz 02:16, 27 June 2007 (UTC)
 * Stigmergic Optimization, in Abraham, Grosan and Ramos (Eds.), Studies in Computational Intelligence, Springer Verlag, Germany, pages 300, ISBN: 3-540-34689-9, 2006.
 * Swarm Intelligence in Data Mining, by A. Abraham, Crina Grosan and Vitorino Ramos (Eds.), Studies in Computational Intelligence, Springer Verlag, Germany, pages 270, ISBN: 3-540-34955-3, 2006 (Preface and foreword by James Kennedy]).

I aggree with Ronz, "orphen" references do not bring any interest, just more noise--Jbw2 18:28, 28 June 2007 (UTC)

I think we shouldn't remove them if these links could bring some more information (links to be assessed), but I suggest to move them from "reference" section to "External links". --Jbw2 18:35, 28 June 2007 (UTC)
 * The links probably don't meet WP:EL nor WP:SPAM. The books might be listed in a Further reading section though, without the links. --Ronz 19:09, 28 June 2007 (UTC)

Good reference
http://www7.nationalgeographic.com/ngm/0707/feature5/index.html

Mathiastck 21:28, 10 July 2007 (UTC)


 * its been proposed that Flocking (behavior) be merged with this. I had never seen teh term Swarm intelligence used, but saw lots of use of Flocking in animation. Mathiastck 23:38, 10 July 2007 (UTC)
 * That merge proposal at Talk:Flocking (behavior) was in December 2004! I imagine the situation has changed since then... ;) Clearly related, but separate topics. --Quiddity 00:04, 11 July 2007 (UTC)

Movie reference incorrect
I believe Boids was used to generate swarms of penguins for Batman in 1992, so this looks to be incorrect. Any reference?

Metaheuristics, not algorithms
Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO) are not algorithms but metaheuristics. The title of Section 2 should be changed.SDas 04:45, 14 August 2007 (UTC)

Section on Stochastic diffusion search
Stochastic diffusion search is a relatively unknown metaheuristic in comparison to ACO and PSO which have witnessed huge interest recently by theoreticians as well as practising scientists and engineers.

I strongly suggest removing it, and enhancing the section on PSO and ACO instead. SDas 04:46, 14 August 2007 (UTC)


 * If the info it contains is true and relevant to the article, then I see no reason to remove it. The article could state that it is of lesser importance than the other algorithms.  NerdyNSK 03:24, 19 September 2007 (UTC)
 * I agree, provided we have reliable sources and are not giving it undue weight or otherwise treating it in an unbalanced manner. --Ronz 03:30, 19 September 2007 (UTC)

Sdas is correct to highlight that SDS is not as well known as PCO and ACA (but then few of the other swarm metaheuristics listed here are) however he is misinformed if he believes SDS is lacks theoretical foundation; it has a much stronger theoretical description than either ACO and PCO (see references now included in main article).. — Preceding unsigned comment added by Markybish (talk • contribs) 16:54, 18 November 2012 (UTC)

Links
The article could benefit, I think, with some more links. For example, it is necessary to link the word pheromone in the article, because readers may have no idea what this is. Years and names of organisations (eg NASA) should be linked as well. NerdyNSK 22:10, 17 September 2007 (UTC)


 * Yup, I agree. My reversal was due to the fact that some wikilinks led to unrelated pages (like Wang for example), and some referenced obviously trivial subjects. I think we can safely assume that a person reading on Swarm Intelligence is familiar with concepts of nature, point, surface, movie, mathematics, clapping, etc :) On other hand pheromone and, for example, combinatorial are less trivial and therefore are good candidates for linking. Alex Pankratov 22:20, 17 September 2007 (UTC)

Business applications
The article can be improved by discussing the business applications of swarm intelligence and related algorithms (which what I research for my MSc). Anyone objecting this for any reason? NerdyNSK 22:13, 17 September 2007 (UTC)

Jing Wang: Help needed
I need help in researching about Jing Wang and writing an article on him. NerdyNSK 03:21, 19 September 2007 (UTC)

Hive mind
I don't see it in the history... has anyone decided that Orson Scott Card's 1985 book Ender's Game is worth adding to the pop culture heading? The plot includes hive-mind battle tactics by insect-like aliens. Binksternet 07:17, 27 September 2007 (UTC)

Empire of the Ants?
Does H. G. Welles' Empire of the Ants deals with hive intelligence, or was his conceptualization more conventional (and thus not relevant to this article)? I also recall another SF short story involving an army ant swarm that developed intelligence enough to grains of gold and rare Amazon butterflies for cattle; I can't recall the title or the author. Any thoughts? rewinn (talk) 18:19, 1 February 2009 (UTC)

Next World
In Discovery Channel "Next World" Program, it was discussed that Grasshopper swarming may be used for road traffic. It is also mentioned that cars can be joined together like trains (while moving), so the wind resistance is less etc.

If anyone know more on these, please add a comment.Photnart (talk) 04:12, 11 October 2009 (UTC).

Artificial or Natural?
There seems to be a contradiction in the first two paragraphs:

"Swarm intelligence (SI) is a type of artificial intelligence ..."

"Natural examples of SI include ant colonies, ..."

If SI is defined to be a type of AI, then the natural examples aren't really examples of SI. Either the definition needs to be broadened to include natural phenomena, or else the natural phenomena need to be presented differently, maybe SI is "modeled on" them. 66.245.43.17 (talk) 19:57, 26 October 2009 (UTC)

Compare the scholarpedia.org article, which begins, "Swarm intelligence is the discipline that deals with natural and artificial systems composed of many individuals that coordinate using decentralized control and self-organization." 66.245.43.17 (talk) 20:07, 26 October 2009 (UTC)

New algorithms
Hi, I have a concern. There are a number of algorithms mentioned in the "Example algorithms" section. However, they are not well-known (compared to PSO and ACO) and never have been used for solving real-world problems. There is no trace of them in the literature. For example, "River formation dynamics" have been used ONLY by its original author and no other researcher have used this algorithm. The "Stochastic diffusion search" algorithm has NO reference at all! The two algorithms "Intelligent Water Drops" and "Charged System Search" have been only used by the original authors! I do not think they qualify to appear on this page unless they are verified by independent researchers and can show their merits in solving real problems. Please let me know what you think. Thanks. - ALI REZA (talk) 13:56, 3 February 2011 (UTC)
 * I agree to an extent. The three most important examples (ACO, PSO and SDS) should be given more emphasis while the other examples should be pushed down a little. However, these other examples have been peer reviewed and published, which I believe gives them enough credibility for at least a passing mention. Might I suggest that under examples we have 4 subheadings, ACO, PSO, SDS and Recent Development (or words to that effect). Under the latter paragraph, the explanations should be brief, and for now should omit comments on their effectiveness, at least until other authors publish work on these techniques. (Some of the existing wording seems a little self-promoting.) Jr271 (talk) 12:52, 30 March 2011 (UTC)

Only one paper of the intelligent Water Drops(IWD) algorithm has been cited so far by 58 other papers and the above statement is false and out-of-dated. — Preceding unsigned comment added by 2.177.99.223 (talk) 19:41, 10 April 2013 (UTC)

DSA and BSA algorithms
The following discussion has been transferred to here from a user talk page.

Dear Sir/Madam,

As you know DSA and BSA links are under an edit-war. Since these algorithms have been proposed recently, it may be thought that they are not well known by the related community. As you can easily check, both DSA and BSA are the 4th most-downloaded papers in the journals they have been published, for several months. The manuscript of DSA has been downloaded more than 4000 times. In my opinion, Wikipedia must not prevent the accession of readers to the recently developed high-performance algorithms such as DSA and BSA. Otherwise optimization community will never be able to develop. Therefore I request you not to edit-war against DSA and BSA links. Analogical-Swarm intelligence models of DSA and BSA can be found in related papers.

Regards, Mehmet Emre Phd.Student — Preceding unsigned comment added by 88.228.174.93 (talk) 20:34, 19 August 2013 (UTC)


 * Hi Mehmet. DSA and BSA are new algorithms, and enough time hasn't passed to say whether they will become notable or not. To date they have been mentioned in only one review, where they were mentioned in passing but were not evaluated. According to Google Scholar the paper you cited as a seminal paper for the BSA has one citation while the paper you cite as a seminal paper for the DSA has nine citations. This is a marked contrast to citations for more notable algorithms. An ACO seminal paper has 2,148 citations, a PSO seminal paper has 6,259 citations, and an SPP seminal paper has 2,414 citations. Time will tell whether DSA and BSA are going to join this company, but that time doesn't seem to have arrived yet. Regards. --Epipelagic (talk) 06:41, 20 August 2013 (UTC)