A. Fukunaga. Restart scheduling for genetic algorithms. In Thomas Back, editor, Genetic Algorithms: Proceedings of the Seventh International Conference, 1997.  Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
When Short Runs Beat Long Runs - Sean Luke George (2001) (3 citations) (Correct)
....expected to move towards the optimum, either in solution space or in fitness space (Back, 1996) Since different population sizes can be considered different techniques, this analysis can shed light on resource allocation issues. One area which directly tackles resource allocation is scheduling (Fukunaga, 1997). A schedule is a plan to perform n runs each l generations long. The idea is to come up with a schedule which best utilizes available resources, based on past knowledge about the algorithm built up in a database. Typically this knowledge is derived from previous applications of the algorithm to ....
....The idea is to come up with a schedule which best utilizes available resources, based on past knowledge about the algorithm built up in a database. Typically this knowledge is derived from previous applications of the algorithm to various problem domains different from the present application. (Fukunaga, 1997) argues that previous problem domains are a valid predictor of performance curves in new domains, for genetic algorithms at least. Outside of evolutionary computation, there is considerable interest in restart methods for global optimization. For difficult problems where one expects to perform ....
A. Fukunaga. Restart scheduling for genetic algorithms. In Thomas Back, editor, Genetic Algorithms: Proceedings of the Seventh International Conference, 1997.
Towards a Rational Methodology for Using Evolutionary Search.. - Sharpe (2000) (Correct)
....problems then HG RWP could perform no better than random search, but we have no need to get concerned by the full wrath of NFL as we are only interested in HG RWP s performance averaged over all possible RWPs. One of the biggest hurdles facing such a HG RWP is the so called re start problem (Fukunaga, 1998). Some RWPs have many long paths that lead from anywhere to the best solution. To search these efficiently you must use an ESA that has the patience to climb such a long path. Other RWPs have a few short paths to optimal solutions, paths that can only be reached from a few start locations. To be ....
....reached from a few start locations. To be successful on these RWPs one needs to use an ESA that re starts a lot of times trying to find one of these paths. Once it hits a path it will quickly find the optimum. These two extremes form the ends of the spectrum of re start schedules (Goldberg, 1999; Fukunaga, 1998) for efficient search (figure 4.5) The problem is how do you know when to re start on a new search space Many, including myself, believe that such problems will ultimately lead to a No Free Lunch theorem for real world problems, NFL RWP . Unfortunately, as mentioned before, the RWPs are not yet ....
Fukunaga, A. (1998). Restart scheduling for genetic algorithms. In Proceedings of the Seventh International Conference on Genetic Algorithms 1998.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC