Vidal's library
Title: No Free Lunch Theorems for Search
Author: David H. Wolpert and William G. Macready
Institution: Santa Fe Institute
Year: 1995
Abstract: We show that all algorithms that search for an extremum of a cost function perform exactly the same, when averaged over all possible cost functions In particular, if algorithm A outperforms algorithm B on some cost functions then loosely speaking there must exist exactly as many other functions where B outperforms A Starting from this we analyze a number of the other a priori characteristics of the search problem, like its geometry and its informationtheoretic aspects This analysis allows us to derive mathematical benchmarks for assessing a particular search algorithms performance We also investigate minimax aspects of the search problem, the validity of using characteristics of a partial search over a cost function to predict future behavior of the search algorithm on that cost function, and timevarying cost functions We conclude with some discussion of the justiability of biologicallyinspired search methods

Cited by 353  -  Google Scholar

@TechReport{wolpert95a,
  author =	 {David H. Wolpert and William G. Macready},
  title =	 {No Free Lunch Theorems for Search},
  institution =	 {Santa Fe Institute},
  year =	 1995,
  abstract =	 {We show that all algorithms that search for an
                  extremum of a cost function perform exactly the
                  same, when averaged over all possible cost functions
                  In particular, if algorithm A outperforms algorithm
                  B on some cost functions then loosely speaking there
                  must exist exactly as many other functions where B
                  outperforms A Starting from this we analyze a number
                  of the other a priori characteristics of the search
                  problem, like its geometry and its
                  informationtheoretic aspects This analysis allows us
                  to derive mathematical benchmarks for assessing a
                  particular search algorithms performance We also
                  investigate minimax aspects of the search problem,
                  the validity of using characteristics of a partial
                  search over a cost function to predict future
                  behavior of the search algorithm on that cost
                  function, and timevarying cost functions We conclude
                  with some discussion of the justiability of
                  biologicallyinspired search methods},
  cluster = 	 {15851006413928005146},
  url = 	 {http://jmvidal.cse.sc.edu/library/wolpert95a.pdf}
}
Last modified: Wed Mar 9 10:14:04 EST 2011