Wikipedia

Wikipedia:Articles for deletion/Logic and the mind

< Wikipedia:Articles for deletion
The following discussion is an archived debate of the proposed deletion of the article below. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.

The result was delete. per WP:OR, WP:ESSAY, WP:FRINGE l'aquatique || talk 01:25, 7 December 2008 (UTC)[reply]

Perlovsky's theory of logic and the mind

edit
Perlovsky's theory of logic and the mind (edit | talk | history | protect | delete | links | watch | logs | views) (delete) – (View log)

Talk page says the text is copied from http://www.leonid-perlovsky.com with permission. This may be original research. Like Wikipedia:Articles for deletion/Knowledge instinct (2nd nomination) this appears to be a one-man theory with very little independent support.

I am also nominating the following article for the same reason:

McWomble (talk) 13:52, 23 October 2008 (UTC)[reply]

  • Keep or at least partial merge parts of Logic and the mind with psychology of reasoning. At least parts of this article could perhaps be profitably merged into our too brief stub on the psychology of reasoning. In fact, I might be tempted to prefer this title over "psychology of reasoning" for the content there, although perhaps logic and psychology might be the best title. The interface between formal logic and actual human reasoning processes is a vital philosophical subject, and our coverage seems scattershot. The historical account given here seems reasonably mainstream, accurate, and valuable to me.

    Neural modeling fields seems to rely much more strongly on Leonid Perlovsky's own thought, and frankly I found it much rougher going. Dr. Perlovsky does seem to be a respected academic, though, and there doesn't seem to be any commercial conflict of interest going on, so I say weak keep to that one. - Smerdis of Tlön (talk) 14:15, 23 October 2008 (UTC)[reply]
  • Comment on NMF. Regardless of what happens with the Logic and the mind, I would like to make sure that the Neural modeling fields article can be kept. I have changed it by removing more controversial claims. Basically, this article is about the mathematics behind the dynamic logic theory, which is a machine learning technique, used by AI researchers, there are many publications, and it definitely deserves to be on Wikipedia. —Preceding unsigned comment added by Romanilin (talk • contribs) 19:15, 26 October 2008 (UTC)[reply]
We need some references to "neural modeling fields" that aren't from Perlovsky. Also, the phrase is a neologism. This seems to be a multi-stage neural net, an idea that dates back to the 1960s, (see Perceptron) but by using nonstandard terminology, it's made to look like a new idea from Perlovsky.--John Nagle (talk) 03:31, 27 October 2008 (UTC)[reply]
OK, will provide the references, give me till tomorrow don't have time today to work on this.Romanilin (talk) 18:36, 28 October 2008 (UTC)[reply]
I updated the references to include several not by Perlovsky. This theory has been referred to by several names, which I added to introduction. In the book by Perlovsky (2001) where he describes the NMF, he calls it Modeling Fields Theory. Regardless, this is a valid NEW idea, that has been used by researchers, and it is NOT the same as multi-stage neural network. Romanilin (talk) 04:37, 29 October 2008 (UTC)[reply]
Then we ought to have an explanation of how they differ - sourced, and preferably not from Perlovsky. Septentrionalis PMAnderson 16:19, 29 October 2008 (UTC)[reply]
I've been looking at the new references. Deming is associated with Perlovsky; they've co-authored papers, so he's not an independent source. I did find a thesis that referenced MLANS [1], but only to reject that approach. Perlovsky is also associated with Nichols Research, referenced in the "MDA Technologies" reference. [2]. After a few mergers and spinoffs, the company "Torch Concepts" emerged with the technology.[3] They were later involved with the JetBlue data mining privacy scandal.[4], but that seems to be unrelated to MLANS. I'm still not seeing anything by an unaffiliated third party. I did find a mention of MLANS and Torch Concepts [5], but it credits Prof. Keinosuke Fukunaga of Purdue University with the technology. At best, MLANS seems to deserve a minor mention in some article on neural nets or a related article. --John Nagle (talk) 17:35, 29 October 2008 (UTC)[reply]
For one thing, Torch does not say that Dr. Burdick came up with MLANS, it says he directed several projects for ARPA and NASA involving the application of MLANS. Now, the name and the idea for this theory do come from Perlovsky, there is not much we can do about it. I thought all we needed was proof that this is used by other researchers. I have been to conferences where people presented on this. True a lot of them are somehow associated with Perlovsky, but how can they not be, the theory is only 20 years old, Perlovsky is still doing active research.Romanilin (talk) 20:53, 29 October 2008 (UTC)[reply]
  • Comment What this subject really seems to deserve is a brief entry in Neural network#Learning algorithms. That's where the various algorithms for training neural nets are covered, and that's where Perlovsky's scheme fits into Wikipedia, if anywhere. It's being presented here as a standalone theory, with few ties to existing work and claims that it's a significant breakthrough, which makes it look WP:FRINGE. As Neural network puts it, "There are many algorithms for training neural networks; most of them can be viewed as a straightforward application of optimization theory and statistical estimation." --John Nagle (talk) 18:21, 30 October 2008 (UTC)[reply]
This is NOT a neural network, this IS a stand alone theory. Could we PLEASE ask an opinion of somebody who IS in the field of Computational Intelligence? This is a model based framework, and neural networks are NOT model based systems. Neural networks consist of neurons (simple processing elements) and weights. NMF system consists of parameterized models, arbitrarily complex. Yes it can be visualized as a neural network but it is not. However, even if it were, not all neural networks are located in one article. For example, Adaptive resonance theory is a neural network architecture that has its own entry. And it is just a type of Neural network. Romanilin (talk) 22:35, 30 October 2008 (UTC)[reply]
I just thought that the article could use a good illustration, I will add it and hopefully clarify the structure and the difference from the neural network.Romanilin (talk) 00:11, 31 October 2008 (UTC)[reply]
By the way, I want to make it clear that I am only arguing about the Neural modeling fields article. The other article Logic and the mind is more controversial and since many people object I am OK with deleting it and reworking it later in a different form or as part of another entry as suggested.Romanilin (talk) 23:20, 30 October 2008 (UTC)[reply]
One of the references even says it's a neural network approach: "The main component of the approach is the maximum likelihood adaptive neural system (MLANS), which is a model-based neural network combining the adaptivity of a neural network with the a priori knowledge of signal models. "[6]. "Computational intelligence" is what used to be called "neural networks" or "connectionism". I don't use that stuff much, but I do have a MSCS degree from Stanford, once took "Epistemological Problems in Artificial Intelligence" from McCarthy, hold some patents in the area, and ran a DARPA Grand Challenge team, so I'm reasonably familiar with the field. This stuff just isn't that novel. Model-based systems have been tried before, usually in the field of adaptive model-based control. Many, many schemes for tuning neural nets have been tried. It's hard to tell where this stuff fits, though, because of the nonstandard terminology, the lack of references to related work, and the general weirdness of the material. I'm thinking WP:FRINGE here. For an example of a similar fringe theory, see [7]. We need more on this subject written independently of Perlovsky. --John Nagle (talk) 02:18, 31 October 2008 (UTC)[reply]

Comment from the author It looks like there are two major objections. Here is my response again to both.

1. "There is not enough support except from the author of the theory"

A. Perlovsky himself is a respected scientist. He wrote a book, many book chapters, hundreds of publications. He received a McLucas Basic Research Award from the US Air Force: http://www.af.mil/news/story.asp?id=123071858

B. NMF theory is described in his 2001 book. The book has good reviews, see Amazon web site.

http://www.amazon.com/Neural-Networks-Intellect-Model-Based-Concepts/dp/0195111621/ref=sr_1_2?ie=UTF8&s=books&qid=1225632930&sr=1-2

C. He wrote several book chapters on NMF

D. There are many publications that describe application of NMF. For example, this paper speaks of 20db (100 times) improvement of tracking in clutter. Ground moving target indication is a difficul problem and the improvement is simply huge. The paper is published in IEEE transactions on neural networks. http://www.leonid-perlovsky.com/perlovsky-TNN06-L487-final2.pdf

E. There are references on the internet to NMF as basis for grants, research proposals etc. For example:

http://www.mda.mil/mdalink/pdf/bneu.pdf

http://www.tech.plym.ac.uk/soc/research/ABC/

http://spie.org/x648.xml?product_id=540989

I think this is hard to claim that there is no link to existing work. If people who use NMF know Perlovsky and co-author with him, that is because the neural networks community is not very big and it is a young field. People working with similar technologies usually collaborate. True, there is no separate book not written by Perlovsky on NMF, but this cannot be a criteria for deleting the page, given all the other references. I also don't think that the NMF article is trying to artificially inflate the importance of NMF, it simply describes what it is mathematically/algorithmically and gives the phycological interpretation.


2. "This is just a regular neural network disguised in different terminology"

Yes the word "neural network" is in the refences. However in order to claim that this is nothing new, simple word search is not enough. With all respect to John Nagle, his main area does not seem to be in neural networks. Romanilin (talk) 13:37, 2 November 2008 (UTC)[reply]

Q: Are you talking about Logic and the mind or Neural modeling fields? This page is for discussing the former. If we don't stick to that topic, it gets much harder for anyone else to follow the discussion. — Carl (CBM · talk) 14:33, 2 November 2008 (UTC)[reply]
A. This is only about Neural modeling fields. Somebody nominated both pages on this discussion, but at this point I am only talking about the second one.
I have the same question, which of the two articles is this about? —Preceding unsigned comment added by Romanilin (talk • contribs) 21:53, 2 November 2008 (UTC)[reply]
Relisted to generate a more thorough discussion so that consensus may be reached.
Please add new comments below this notice. Thanks, McWomble (talk) 09:53, 4 November 2008 (UTC)[reply]
  • Huh? Delete it merely because it's a fringe theory? There's a consensus to delete those??? I don't think that's what WP:FRINGE says: It says this:
This guideline establishes which fringe theories and opinions may be included in Wikipedia, and to a certain extent how articles about them should approach their subjects.
It says it's about which ones should be included. It's not about a guideline saying to delete them all. I think it's got to have something to do with notability. Michael Hardy (talk) 23:33, 6 November 2008 (UTC)[reply]
The requirements are, in a nutshell.
  • In order to be notable enough to appear in Wikipedia, a fringe idea should be referenced extensively, and in a serious manner, in at least one major publication, or by a notable group or individual that is independent of the theory.
  • Even debunking or disparaging references are adequate, as they establish the notability of the theory outside of its group of adherents.
This seems reasonable; we should have articles on Velikovsky or on the New Chronology or on John Cleves Symmes. But I don't see that either of these is satisfied here. Septentrionalis PMAnderson 23:51, 6 November 2008 (UTC)[reply]
Many other than Velikovsky and his associates have written about Velikovsky's fringe theories, but few other than Perlovsky and his associates have written about Perlovsky's fringe theories. That's the notability distinction. --John Nagle (talk) 20:30, 7 November 2008 (UTC)[reply]
I agree with John Nagle; I apologize for having been unclear. Septentrionalis PMAnderson 21:23, 7 November 2008 (UTC)[reply]
  • Delete. No evidence that this theory is notable. All of the references are either to articles by the theory's originator or to historical writings -- none indicate that that the theory is considered notable by anyone other than Perlovsky. The whole article looks like OR, but if there's anything salvageable, it could be merged to the Leonid Perlovsky article. Klausness (talk) 20:03, 8 November 2008 (UTC)[reply]
  • This is NOT fringe. And not OR, the article Neural modeling fields is based on published work. Come on, this is getting ridiculous. Here is definition from wiki: "Fringe science is scientific inquiry in an established field of study which departs significantly from mainstream or orthodox theories, and is classified in the "fringes" of a credible mainstream academic discipline. Mainstream scientists typically regard fringe concepts as highly speculative or weakly confirmed". There is a book and many publications. Certainly Oxford monograph is sufficient to establish a mainstream. There are many publications in serious journals. US Air Force Basic Research Award is not given for fringe theories, International Neural Networks Society Gabor Award is not given for fringe theories. Look at Perlovsky web site. Not to mention that Computational intelligence is a young field of study, so what we are not going to put anything on the wiki until there is more than one book about it? —Preceding unsigned comment added by Romanilin (talk • contribs) 19:50, 9 November 2008 (UTC)[reply]
  • Delete. This article is fundamentally flawed and nothing could be done to make it okay. Statements like "Aristotle invented logic" where the ref is from Aristotle are deeply misleading. Does Perlovsky think Aristotle invented logic or does Aristotle think Aristotle invented logic? The article argues a point (rather feebly if you ask me) but that is not an encyclopedia entry's role. There is too much synthesis and original research here. And it couldn't be otherwise! xschm (talk) 21:24, 11 November 2008 (UTC)[reply]
  • What is the relationship between the claim that Aristotle invented logic and Perlovsky's theory of logic and the mind? This article gives a cursory (and wildly incomplete) history of theories of logic. But it doesn't relate them to the subject of the article. It seems like it aims to be an essay arguing that Perlovsky is the apotheosis of this grand tradition. It fails to make a convincing argument and such an argument has no place in an encyclopedia. If that argument exists elsewhere, it could be documented here, but it seems far from clear that is the case. xschm (talk) 01:10, 12 November 2008 (UTC)[reply]
  • Delete. This appears to be a personal essay of the type described in WP:NOT. The contents are not *about* Perlovsky's theories and how they might have been received by reliable third parties, they *are* Perlovsky's theories. The map should not be the territory, but in this case it is. —David Eppstein (talk) 05:42, 12 November 2008 (UTC)[reply]
Relisted so that consensus may be reached.
Please add new comments below this notice. Thanks, xschm (talk) 21:27, 17 November 2008 (UTC)[reply]
  • Comment he himself is notable, whether or not mainstream, on the basis of OUP publishing of one of his books. That does not make every one of his theories, or any of them, separately notable. One article is sufficient. most of the material here is unencyclopedic summary and argumentation, so I don't see how there's anything appropriate for merging. This is a clear attempt to use Wikipedia for promotional purposes. DGG (talk) 04:45, 19 November 2008 (UTC)[reply]
Yes, exactly that. Well said. Guy (Help!) 01:29, 20 November 2008 (UTC)[reply]
Agreed. We're not proposing to delete the Leonid Perlovsky article here, just the "spinoff articles". --John Nagle (talk) 05:09, 20 November 2008 (UTC)[reply]
  • Comment I think there's consensus for deleting Perlovsky's theory of logic and the mind. That's just a badly written history of logic with a link to Perlovsky's stuff at the end. It's still not clear what to do about Neural modeling fields, which is more like a technical paper. If that's kept, it will need a major rewrite, which is going to be a tough job. Maybe trim it down to a brief note on what the subject is about, with a few links to papers, other work, and related neural net articles. --John Nagle (talk) 17:46, 22 November 2008 (UTC)[reply]
The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia:Articles_for_deletion/Logic_and_the_mind&oldid=1180006149"