The recent application of Machine Learning techniques to the Answer Set Programming (ASP) field proved to be effective. In particular, the multi-engine ASP solver ME-ASP is efficient: it is able to solve more instance...
详细信息
The recent application of Machine Learning techniques to the Answer Set Programming (ASP) field proved to be effective. In particular, the multi-engine ASP solver ME-ASP is efficient: it is able to solve more instances than any other ASP system that participated to the 3rd ASP Competition on the 'System Track' benchmarks. In the ME-ASP approach, classification methods inductively learn offline algorithmselection policies starting from both a set of features of instances in a training set, and the solvers performance on such instances. In this article we present an improvement to the multi-engine framework of ME-ASP, in which we add the capability of updating the learned policies when the original approach fails to give good predictions. An experimental analysis, conducted on training and test sets of ground instances obtained from the ones submitted to the 'System Track' of the 3rd ASP Competition, shows that the policy adaptation improves the performance of ME-ASP when applied to test sets containing domains of instances that were not considered for training.
Answer-set programming (ASP) is a truly declarative programming paradigm proposed in the area of non-monotonic reasoning and logic programming, which has been recently employed in many applications. The development of...
详细信息
Answer-set programming (ASP) is a truly declarative programming paradigm proposed in the area of non-monotonic reasoning and logic programming, which has been recently employed in many applications. The development of efficient ASP systems is, thus, crucial. Having in mind the task of improving the solving methods for ASP, there are two usual ways to reach this goal: (i) extending state-of-the-art techniques and ASP solvers or (ii) designing a new ASP solver from scratch. An alternative to these trends is to build on top of state-of-the-art solvers, and to apply machine learning techniques for choosing automatically the "best" available solver on a per-instance basis. In this paper, we pursue this latter direction. We first define a set of cheap-to-compute syntactic features that characterize several aspects of ASP programs. Then, we apply classification methods that, given the features of the instances in a training set and the solvers' performance on these instances, inductively learn algorithmselection strategies to be applied to a test set. We report the results of a number of experiments considering solvers and different training and test sets of instances taken from the ones submitted to the "System Track" of the Third ASP Competition. Our analysis shows that by applying machine learning techniques to ASP solving, it is possible to obtain very robust performance: our approach can solve more instances compared with any solver that entered the Third ASP Competition.
暂无评论