Diversify Intensification Phases in Local Search for SAT with a New Probability Distribution
View/ Open
Author(s)
Duong, TT
Pham, DN
Sattar, A
Griffith University Author(s)
Year published
2013
Metadata
Show full item recordAbstract
A key challenge in developing efficient local search solvers is to intelligently balance diversification and intensification. This study pro- poses a heuristic that integrates a new dynamic scoring function and two different diversification criteria: variable weights and stagnation weights. Our new dynamic scoring function is formulated to enhance the diversifi- cation capability in intensification phases using a user-defined diversifica- tion parameter. The formulation of the new scoring function is based on a probability distribution to adjust the selecting priorities of the selection between greediness on scores ...
View more >A key challenge in developing efficient local search solvers is to intelligently balance diversification and intensification. This study pro- poses a heuristic that integrates a new dynamic scoring function and two different diversification criteria: variable weights and stagnation weights. Our new dynamic scoring function is formulated to enhance the diversifi- cation capability in intensification phases using a user-defined diversifica- tion parameter. The formulation of the new scoring function is based on a probability distribution to adjust the selecting priorities of the selection between greediness on scores and diversification on variable properties. The probability distribution of variables on greediness is constructed to guarantee the synchronization between the probability distribution func- tions and score values. Additionally, the new dynamic scoring function is integrated with the two diversification criteria. The experiments show that the new heuristic is efficient on verification benchmark, crafted and random instances.
View less >
View more >A key challenge in developing efficient local search solvers is to intelligently balance diversification and intensification. This study pro- poses a heuristic that integrates a new dynamic scoring function and two different diversification criteria: variable weights and stagnation weights. Our new dynamic scoring function is formulated to enhance the diversifi- cation capability in intensification phases using a user-defined diversifica- tion parameter. The formulation of the new scoring function is based on a probability distribution to adjust the selecting priorities of the selection between greediness on scores and diversification on variable properties. The probability distribution of variables on greediness is constructed to guarantee the synchronization between the probability distribution func- tions and score values. Additionally, the new dynamic scoring function is integrated with the two diversification criteria. The experiments show that the new heuristic is efficient on verification benchmark, crafted and random instances.
View less >
Conference Title
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume
8272 LNAI
Publisher URI
Copyright Statement
© 2013 Springer Berlin/Heidelberg. This is the author-manuscript version of this paper. Reproduced in accordance with the copyright policy of the publisher.The original publication is available at www.springerlink.com
Subject
Artificial intelligence not elsewhere classified