Towards a formal specification of local search neighborhoods from a constraint satisfaction problem structure
Loading...
Date
Journal Title
Journal ISSN
Volume Title
Publisher
ACM
Abstract
Neighborhood operators play a crucial role in defining effective Local Search solvers, allowing one to limit the explored search space and prune the fitness landscape. Still, there is no accepted formal representation of such operators: they are usually modeled as algorithms in procedural language, lacking in compositionality and readability. In this paper we outline a new formalization capable of representing several neighborhood operators eschewing their coding in a full Turing complete language. The expressiveness of our proposal stems from a rich problem representation, as used in Constraint Programming models. We compare our system to competing approaches and show a clear increment in expressiveness.
Description
Keywords
Citation
Mateusz Ślażyński, Salvador Abreu, and Grzegorz J. Nalepa. 2019. Towards a formal specification of local search neighborhoods from a constraint satisfaction problem structure. In Proceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO '19), Manuel López-Ibáñez (Ed.). ACM, New York, NY, USA, 137-138. DOI: https://doi.org/10.1145/3319619.3321968