Notices
Results 1 to 4 of 4

Thread: Two level search

  1. #1 Two level search 
    New Member
    Join Date
    Jul 2010
    Posts
    2
    Hi!
    This is my first post. I need help.

    Local search (hill climbing, for example) search solution space for the best solution. Is there any known combination of two cascade search "engines". First one, inner, search what we really need (we will search for this anyway). And the second search engine search over the results from that inner search, by changing various parameters, maybe even change applied search algorithm of the inner search engine. So the outer search is used as kind of benchmark (but progressively leading to a best combination for a given problem set).

    Firstly, i named it 'meta search', but it is already used for something quite different (aggregation of results from many search engines).
    Then I have tried 'hyper search', which is used too.

    multilevel, cascade... all seem to be used for something else.

    As I know, principle I have just described, is not known in the CS. I would appreciate it if someone could either send me a link toward existing principle, or give me a suggestion for naming it.


    Reply With Quote  
     

  2.  
     

  3. #2  
    Forum Freshman
    Join Date
    Feb 2010
    Posts
    84
    The IEEE "Computational Intelligence" journal for May 2010 has a buncha articles on "memetic computing" which seems to be a fancy word for using "whatever works" techniques, mostly for searching. For some reason, maybe related to why CSers use "ontology" to mean "category", they chose to use the word "meme" to represent different solution approaches. It does sound more important, doesn't it?

    I guess a simple example would be to use a genetic algorithm to find areas of likely interest in a fitness landscape, and then switch to a hill-climb to maximize. Or perhaps modify the genetic parameters significantly to narrow the search.


    Reply With Quote  
     

  4. #3  
    New Member
    Join Date
    Jul 2010
    Posts
    2
    Thanks. I'll try to look at that journal. Just saw Memetic Computing, Springer's journal.
    Also, you are right that genetic algorithm may be used to quick sweep for promising areas. But unfortunately, I am currently stuck with local search engines. Maybe later. I wonder, since there could be such imaginative combinations of neural nets/genetic/fuzzy/search/... why isn't there any generic naming convention.

    I would like to find some easy to read survey about those interesting combinations and guidance for their recommended usage. Any advice?
    Reply With Quote  
     

  5. #4  
    Forum Freshman
    Join Date
    Feb 2010
    Posts
    84
    I was surprised to find that it's such a big deal, so I don't know much about it. The wiki page might be the place to start: http://en.wikipedia.org/wiki/Memetic_algorithm
    Reply With Quote  
     

Bookmarks
Bookmarks
Posting Permissions
  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •