S. Rana and D. Whitley. Genetic algorithm behavior in the MAXSAT domain. In Proceedings of the 5th Conference on Parallel Problem Solving from Nature, pages 784-794, 1998. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
On the Role of Deception in Second Order Functions - Schoofs, Naudts (Correct)
....(GAs) on second order functions. While this class of optimization problems is NP hard, many of its instances are relatively easy to solve. Nevertheless, a GA has trouble nding a solution quickly on these easy problems. Our aim is to determine which forces slow down the GA; earlier research [6] on the MAX3SAT problem, which contains third order interactions, indicates that deception and a at search landscape may be the main cause. As a representative class of second order functions we use random binary constraint satisfaction problems (CSPs) for which well studied parameterized ....
....The parameter p has the same starting point and the same step size as in 1(a) but its maximum value is 0.17. In both cases the average is taken over 100 instances. We observe that the rst order Walsh coecients are larger than the second order Walsh coecients. Note that this is also observed in [6] where the Walsh coecients of MAX3SAT instances are compared. It explains why algorithms based on rst order information do relatively well on these instances. But they will not consistently lead to an optimum, because there may be second order coecients which contradict the rst order schemas. ....
S. Rana and D. Whitley. Genetic algorithm behavior in the MAXSAT domain. In Proceedings of the 5th Conference on Parallel Problem Solving from Nature, pages 784-794, 1998.
Estimating the Significant Non-Linearities in the Genome.. - Thierens (1999) (1 citation) (Correct)
....: However one should not conclude from this that once we know the signi cant Walsh coecients of functions with a modest amount of non zero Walsh coe cients then this should lead to an ecient GA to optimise the problem. Indeed recent work of Rana and her coworkers (Rana, Heckendorn, Withley, 1998) showed that Boolean Satis ability problems evaluated as MAXSAT functions have a highly restricted set of nonzero Walsh coecients and those coecients can be computed in linear time with respect to the number of clauses. Since MAXSAT is NP Complete the GA cannot be expected to nd the global ....
Rana, S, and Whitley, D. (1998). Genetic Algorithm Behavior in the MAXSAT Domain. Parallel Problem Solving from Nature - PPSN V, Lecture Notes in Computer Science, Vol.1498, p.785-794.
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