Needles in Haystacks Are Not Hard to Find with Neutrality YU T., MILLER J. Abstract: We propose building neutral networks in needle-in-haystack fitness landscapes to assist an evolutionary algorithm to perform search. The experimental results on four different problems show that this approach improves the search success rates in most cases. In situations where neutral networks do not give performance improvement, no impairment occurs either. We also tested a hypothesis proposed in our previous work. The results support the hypothesis: when the ratio of adaptive/neutral mutations during neutral walk is close to that of fitness improvement step, the evolutionary search has a high success rate. Moreover, the ratio magnitudes indicate that more neutral mutations (than adaptive mutations) are required for the algorithms to find a solution in this type of search space. Session: EuroGP Session 2: Experimental & theoretical analysis: April 3, 1400-1605 @inproceedings(yu:2002:EuroGP, title = {Needles in Haystacks Are Not Hard to Find with Neutrality}, author = {Tina Yu and Julian Miller}, editor = {Evelyne Lutton and James A. Foster and Julian Miller and Conor Ryan and Andrea G. B. Tettamanzi}, booktitle = {Proceedings of the 4th European Conference on Genetic Programming, EuroGP 2002}, volume = {2278}, series = {LNCS}, pages = {16--27}, publisher = {Springer-Verlag}, address = {Kinsale, Ireland}, publisher_address = {Berlin}, month = {3-5 April}, year = 2002, keywords = {genetic algorithms, genetic programming}, abstract = {We propose building neutral networks in needle-in-haystack fitness landscapes to assist an evolutionary algorithm to perform search. The experimental results on four different problems show that this approach improves the search success rates in most cases. In situations where neutral networks do not give performance improvement, no impairment occurs either. We also tested a hypothesis proposed in our previous work. The results support the hypothesis: when the ratio of adaptive/neutral mutations during neutral walk is close to that of fitness improvement step, the evolutionary search has a high success rate. Moreover, the ratio magnitudes indicate that more neutral mutations (than adaptive mutations) are required for the algorithms to find a solution in this type of search space. }, notes = {EuroGP'2002, part of lutton:2002:GP}, ) Publisher: Springer-Verlag GmbH ISSN: 0302-9743 Subject: Computer Science Volume 2278 / 2002 Title: Genetic Programming: 5th European Conference, EuroGP 2002, Kinsale, Ireland, April 3-5, 2002. Proceedings Editors: J.A. Foster, E. Lutton, J. Miller, C. Ryan, A.G.B. Tettamanzi (Eds.): Chapter: p. 13 Finding-Needles-in-Haystacks-Is-Not-Hard-with-Neutrality Tina Yu A1 and Julian Miller A2 A1 ChevronTexaco Information Technology Company, San Ramon CA 94583, U.S.A. tiyu@chevrontexaco.com, http://www.improvise.ws A2 School of Computer Science, University of Birmingham, Birmingham B15 2TT, U. K. j.miller@cs.bham.ac.uk, http://www.cs.bham.ac.uk/~jfm Abstract: We propose building neutral networks in needle-in-haystack fitness landscapes to assist an evolutionary algorithm to perform search. The experimental results on four different problems show that this approach improves the search success rates in most cases. In situations where neutral networks do not give performance improvement, no impairment occurs either. We also tested a hypothesis proposed in our previous work. The results support the hypothesis: when the ratio of adaptive/neutral mutations during neutral walk is close to the ratio of adaptive/neutral mutations at the fitness improvement step, the evolutionary search has a high success rate. Moreover, the ratio magnitudes indicate that more neutral mutations (than adaptive mutations) are required for the algorithms to find a solution in this type of search space. -------------------------------------------------------------------------------- Order Confirmation Click the item to access it directly. 1 to 1 of 1 1 . Finding Needles in Haystacks Is Not Hard with Neutrality Type: Individual Article (Electronic Only) First page Previous page Next page Last page This page provides quick access to the items just ordered. These items are also stored in your Order History, along with anything else you may have previously ordered. Your order number is: 2005102121265053 An Invoice is now available to be printed or saved for your records. Frequently asked questions | General information on journals and books c Springer. Part of Springer Science+Business Media | Privacy, Disclaimer, Terms and Conditions, c Copyright Information Remote Address: 82.209.198.29 ? Server: MPWEB20 HTTP User Agent: Mozilla/4.0 (compatible; MSIE 5.01; Windows 98; www.k2pdf.com; FunWebProducts)