No abstract

In this paper we propose a novel approach of rule learning called Relaxed Separate-and- Conquer (RSC): a modification of the standard Separate-and-Conquer (SeCo) methodology that does not require elimination of covered rows. This method can be seen as a generalization of the methods of SeCo and weighted covering that does not suffer from fragmentation. We present an empirical investigation of the proposed RSC approach in the area of Predictive Maintenance (PdM) of complex manufacturing machines, to predict forthcoming failures of these machines. In particular, we use for experiments a real industrial case study of a Continuous Compression Moulding (CCM) machine which manufactures the plastic bottle closure (caps) in the beverage industry. We compare the RSC approach with a Decision Tree (DT) based and SeCo algorithms and demonstrate that RSC significantly outperforms both DT based and SeCo rule learners. We conclude that the proposed RSC approach is promising for PdM guided by rule learning.

The authors wish to make the following corrections to their paper [...]

No abstract

The Weighted Constraint Satisfaction Problem (WCSP) is a popular formalism for encoding instances of hard optimization problems. The common approach to solving WCSPs is Branch-and-Bound (B&B), whose efficiency strongly depends on the method of computing a lower bound (LB) associated with the current node of the search tree. Two of the most important approaches for computing LB include (1) using local inconsistency counts, such as Maintaining Directed Arc-Consistency (MDAC), and (2) Russian Doll Search (RDS).In this paper we present two B&B-based algorithms. The first algorithm extends RDS. The second algorithm combines RDS and MDAC in an adaptive manner. We empirically demonstrate that the WCSP solver combining the above two algorithms outperforms both RDS and MDAC, over all the problem domains and instances we studied. To the best of our knowledge this is the first attempt to combine these two methodologies of computing LB for a B&B-based algorithm.