Jump to content

Talk:Evolutionary algorithm

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

In artificial intelligence

[edit]

"In artificial intelligence, an evolutionary algorithm..." An evolutionary algorithm should not be a subset of artificial intelligence specifically, since you can evolve unintelligent things (ex. theo jansen's mechanical leg for strandbeest). —Preceding unsigned comment added by 68.238.139.213 (talk) 04:37, 30 January 2008 (UTC)[reply]

I think what is meant is that the evolutionary algorithm is itself a simple form of artificial intelligence. The fact that it has many applications, such as refining hull shapes in boat racing, doesn't make it not artificial intelligence. -- Another Stickler (talk) 06:06, 29 October 2008 (UTC)[reply]

Comparison with biological evolution

[edit]

We might want to add a blurb about how the connection with actual biology is more in the way of inspiration and analogy than anything rigorous - as I recall, biological evolution is optimizing against a moving target (an important difference). --Gwern (contribs) 22:17 17 June 2007 (GMT)

If using co-evolution, you may be able to say it's optimizing on a moving fitness landscape. The sorts of selection such as gladitorial tournament selection (you pick a couple of genes, pit them against each other, usually the winner reproduces deleting the loser) makes your fitness depends heavily on what's around you. And yes it's probably not rigorous right now, and people are trying to duplicate the genotype-phenotype distinction in biology and other things. I can't say whether it's completely loosely based right now. Obscurans 22:10, 18 June 2007 (UTC)[reply]
I think it's accurate to say that "an evolutionary algorithm is not intended to be an accurate simulation of an actual biological evolutionary process." We could add something like this. Dcoetzee 07:12, 29 October 2008 (UTC)[reply]

Shouldn't this be part of the series on evolutionary biology? If I knew how to add it to it I would. — Preceding unsigned comment added by 50.170.147.56 (talk) 09:53, 24 August 2015 (UTC)[reply]

[edit]

Dear Oli,

I followed wikipedia's copyright policy, but could not see which part of my revision refringed copyright. It must be very easy for you to spot - i wonder if you'd be kind enough to spare a minute to point out which part to me, please? Appreciated.

The author of the page http://userweb.elec.gla.ac.uk/y/yunli/ga_demo/ga_demo.htm has donated its copyright:

"Copyleft: The text of this webpage is available for modification and reuse under the terms of the Creative Commons Attribution-Sharealike 3.0 Unported License and the GNU Free Documentation License." ieee (talk) 20:08, 17 January 2010 (UTC)[reply]

Make this more accessible

[edit]

I only have one PhD. I have trouble understanding this entire article.

Wikipedia is intended to be accessible to the general public (or at least the typical, well-educated person). This article can only be understood by the contributors to its content, therefore serving no purpose. —Preceding unsigned comment added by 74.138.129.155 (talk) 22:55, 20 July 2010 (UTC)[reply]

In the context of the comments of the above contributor, could a brief explanation of how algorhythms can correspond with biological/environmental models, the relative strengths/possible weaknesses (if any) of the analogy, etc., be added ? — Preceding unsigned comment added by 109.77.99.158 (talk) 21:30, 13 April 2012 (UTC)[reply]

Minor Optimization Methods

[edit]

Some authors of the papers edit these important pages of wikipedia and put their recent methods which are not verified in the literature in these article. For example someone has put his algorithm called cuckoo in all the pages about optimization and evolutionary computation. I think wikipedia is not a tool for advertising methods and doing science marketing. If we are going to list some minor algorithms like these algorithm in the list of important optimization methods, this list should include more that 100 algorithms.

Lets use wikipedia to share the knowledge not to highlight our works. —Preceding unsigned comment added by 74.194.214.87 (talk) 22:54, 23 November 2010 (UTC)[reply]

Questioning the no assumption statement

[edit]

The statement that I am talking about is in the second para. I am finishing up a Ph.D in EC and AI with specific focus on the biases in optimization algorithms. I do not believe evolutionary algorithms make no assumptions or have no biases about the search space. In fact, as per the No free lunch theorem, it would be impossible for an algorithm to outperform the extremely rudimentary random sampling algorithm if the biases are useless. Moreover, only this random sampling is an algorithm with no biases.

The core bias in evolutionary computation is the binary search space which is typical. Without this bias, the search space will simply be a set of numbers with no relationships between them. The binary search space introduces the bias of correlation between fitness and hamming distance, whether true or not. This is why problems for which this is true such as Gen 1-MAX are easy with O(n log n) for EAs (http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.10.5599). Another way of saying this is that the fact that you talk about a fitness landscape at all is a bias. If you truly had no bias, there would not be any landscape to speak of. It would just be a S -> R map. — Preceding unsigned comment added by Gost80 (talkcontribs) 17:33, 11 September 2011 (UTC)[reply]

Moreover, the primary reason EC has enjoyed a lot of success is not just because its base bias is nothing more than a multi-dimensional search space. It is also because it acts as a framework for inserting a large number of biases though custom selection operators and custom mutation operators and representations. Its relative lack of primary bias reduces interference with custom biases allowing it to specialise in a wider variety of fields. But this happens only with the effort of scientists who build in custom biases into the EA though operators and such. — Preceding unsigned comment added by Gost80 (talkcontribs) 19:24, 11 September 2011 (UTC)[reply]

I donit really know what this article is about but it seems like it addresses Evolutionary biology. Soranoch (talk) 19:17, 18 October 2013 (UTC)[reply]

Including Quality Diversity algorithms

[edit]

Quality Diversity (QD) algorithms has been increasingly studied in the past few years. Hence, I added it as a type of evolutionary algorithm, including some prominent works. Another user MrOllie seems to think "Still inappropriate external links, still predatory publisher sourcing". It is very valuable for QD algorithms to be introduced here. If others have better citations to add, I would definitely be glad to change to those! Instead of removing valuable information. Tinkerbrain (talk) 03:28, 28 October 2023 (UTC)[reply]

You shouldn't be including external links in wikipedia mainspace, they can go in an "External Links" section only, see WP:EL. I think the predatory journal that MrOllie is opposed to is Frontiers. They do participate in some predatory practices, although I don't think they can be necessarily be deemed unreliable. In any case, you provided other references, including Nature. I will edit your contribution so that it doesn't include an external link in mainspace. Polyamorph (talk) 08:06, 28 October 2023 (UTC)[reply]
Thank you! Tinkerbrain (talk) 19:36, 28 October 2023 (UTC)[reply]