Jump to content

Talk:Heuristic algorithm

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Virusscanners

[edit]

Should there be a section concerning heuristic scanning in antivirus applications? Or is there already an article on that.. User:Fuzzbox 30 jun 2006

Yes, there is a section of the Antivirus software which describes how heuristics are used to detect suspicious behavior. It links back to this article though, so I'm not sure how to include it. Speaking of which, can anyone explain why there is a "citation needed" tag on the sentence about the use of heuristics in virus scanners? Would linking to the above article be sufficient in place of a citation in this instance? Undisputedloser (talk) 23:55, 18 June 2008 (UTC)[reply]

==

The link on "confidence" near the end of the article references a political stub, when the subject is statistical confidence.

Provably?

[edit]

Well, I really have to admit that I don't know what "provably" (not "probably") means. Perhaps adding a small note to explain what it means, or a link to an article that does, is in order?

--Oskilian

It's the adverb of "provable", which means "there is a proof". Do you not know what a proof is, or did you not understand the flexion? It looks like a perfectly normal word to me, and there's over a million Google hits for it... Unfortunately, I don't know how to avoid this word without losing clarity. --84.59.189.46 19:42, 7 February 2006 (UTC)[reply]
Hmm, I do have to admit I didn't know the flexion and would have probably mistyped it as "proveably" if necessary (And I'm not the alone! google has more than 30000 pages on the word "proveable"!). However, I think that even though the article is properly written, it is not clear for those that don't know the flexion (There's a huge note on the page source explaining that it's supposed to be "provably" and not "probably". I think that it makes my point). To fix this, I've added a link to Proof theory on the word, so people those who don't know the flexion, understand that the word comes from "proof" instead of "probability". But, as a long-term solution, I think that the phrase should be rewritten to something like "finding algorithms with run times that can be proven to be good", which I think is more non-english-native-speakers friendly, without losing any clarity. --Oskilian

two well-defined .. meanings

[edit]

Where are they? It'd be nice to have summary definitions in the intro, so that maybe the intro holds at least a little meaning. 67.161.46.169 01:00, 14 August 2006 (UTC)[reply]

And how!

Wiki heuristics

[edit]

Anybody care to write about Ward Cunningham's WikiWikiWeb project [1] and its methods? And maybe even how the heuristic method can relate to Wikipedia?--Shtove 19:50, 22 April 2007 (UTC)[reply]

If so, then not in this article, which is about technical, well-defined meanings of "heuristic" in computer science. It seems people continuously miss this point, however I am at a loss how to make this any clearer... --Mellum 07:29, 23 April 2007 (UTC)[reply]
Creat a disambiguation header: "This article is about ... For other uses see ..." - but what would you disambiguate it from?--Shtove 18:56, 23 April 2007 (UTC)[reply]
Doh! Already done.--Shtove 09:36, 24 April 2007 (UTC)[reply]

Reference

[edit]

This entry reads like a summary of chapter four of "Artificial Intelligence: A Modern Approach" by Stuart Russell and Peter Norvig, even down to the phrasing of ABSOLVER on page 108 of the second edition of the book. Should this book be referenced?

Anothercodejunkie 02:39, 2 September 2007 (UTC)[reply]

Well, I wrote large parts of this article, and I don't even know that book. But it would certainly be a good reference for the ABSOLVER stuff, so feel free to add it there. --Mellum 12:00, 2 September 2007 (UTC)[reply]

What's the reference to "Pearl (1984)"? Houseofwealth (talk) 23:16, 19 December 2007 (UTC)[reply]

The Pearl reference seems likely to be taken out of context, and its meaning isn't clear. Most heuristics are just things that work in practice, often the result of a human's hunch, not of "intelligent search strategies". If anything, "intelligent search strategies" are the result of heuristics. If "heuristic methods" are different to heuristics, that should be clarified. —Preceding unsigned comment added by 212.44.6.81 (talk) 11:32, 29 January 2009 (UTC)[reply]

Heuristic = Rules of Thumb ?

[edit]

Reading the article, I got the impression it means simplified calculations, perhaps when not enough accurate data exists) that work well enough within certain agreed parameters. The human equivalent seems to be "rule of thumb" which are good enough for non-critical realworld situations. Is this correct ? A couple of real computer examples in layman's terms would improve the article I think. Rcbutcher (talk) 15:13, 10 July 2008 (UTC)[reply]

Along those same lines, is the lead paragraph even correct? Is a heuristic algorithm really "an algorithm that ignores whether the solution to a problem can be proven to be correct"? I thought it was an algorithm that couldn't be proven to give the best answer. (I'm out of my field, here.)--Christopher King (talk) 14:37, 8 January 2009 (UTC)[reply]

Heuristic algorithm is not the same as new algorithm, which proof is not found yet. Sometime some opponent uses the word "heuristic" to note poor mathematical basis for some algorithm, but it is not true usage of this word. Also, some areas of programming can develop algorithm without expenses for its strong proof - for example, game programming area. To optimize development expenses the developers may "ignore whether the solution to a problem can be proven to be correct".--Tim32 (talk) 22:46, 19 January 2009 (UTC)[reply]


propertys of huristics

[edit]

easy to compute: polinomialic in compute time.

admisable, lower bound to the optimal solution. for n is node from N(group of all nodes) and o is optimum value, h is huristic function : N->number so for all n goes h(n)<=o(n)

consistent, for the huristic value from node n is to describe the shortest path to the goal. meaning that for n to the goal is shorter then optimal from n to node m, plus the huristic value of m to the goal. such as: h(n)<=opt(n,m)+h(m)

monotonic, for any node its heuristic cost is higher then its father heuristic value

positive ,in order to prevent negative loops or reaching a state beyond the goal(were goal should always be zero). —Preceding unsigned comment added by 132.72.44.49 (talk) 18:34, 26 January 2010 (UTC)[reply]

Redirected

[edit]

Dear All,

This page has been redirected to metaheuristic which is commonly used in the literature to mean heuristic optimizer, black-box optimizer, etc. The term heuristic algorithm is in fact self-contradictory if you see the dictionary definition: Heuristic: Computers, Mathematics. pertaining to a trial-and-error method of problem solving used when an algorithmic approach is impractical.

Furthermore the page here was very sketchy and the metaheuristic page has just been updated.

I hope you all agree with this change.

Optimering (talk) 13:43, 16 May 2010 (UTC)[reply]