DocumentCode :
2856312
Title :
Towards a Generic CNF Simplifier for Minimising Structured Problem Hardness
Author :
Anbulagan ; Slaney, John
Author_Institution :
NICTA, Australian Nat. Univ., Canberra, ACT, Australia
fYear :
2009
fDate :
2-4 Nov. 2009
Firstpage :
99
Lastpage :
106
Abstract :
CNF simplifiers play a very important role in minimising structured problem hardness. Although they can be used in an in-search process, most of them serve in a pre-search phase and rely on one form or another of resolution. Based on our understanding about problem structure, in the paper, we extend the single pre-search process to a multiple one in order to further simplify the hard structure in a problem. This extension boosts the performance of state-of-the-art clause learning and lookahead based SAT solvers when solving both satisfiable and unsatisfiable instances of many real-world hard combinatorial problems.
Keywords :
computability; learning (artificial intelligence); SAT solvers; conjunctive normal form; generic CNF simplifier; machine learning techniques; real-world hard combinatorial problems; satisfiability; single presearch process; state-of-the-art clause learning; structured problem hardness minimisation; Artificial intelligence; Investments; Large-scale systems; Laser sintering; Learning; Polynomials; Power generation; Random number generation; Stochastic systems; Sugar; CNF Simplifier; Real-world benchmarks; Satisfiability Problem;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Tools with Artificial Intelligence, 2009. ICTAI '09. 21st International Conference on
Conference_Location :
Newark, NJ
ISSN :
1082-3409
Print_ISBN :
978-1-4244-5619-2
Electronic_ISBN :
1082-3409
Type :
conf
DOI :
10.1109/ICTAI.2009.47
Filename :
5365721
Link To Document :
بازگشت