Author:

Abstract:

Data mining (as well as machine learning) are well-established fields of research that are concerned with the analysis of data in order to discover regularities in the form of patterns or functions. Contemporary methods of data mining and machine learning rely heavily on the use of constraints on the patterns or functions of interest. This has resulted in notions of, for instance, constraint-based mining and constrained clustering. Despite the obvious relationships to constraint programming, there has not yet been a lot of work on using constraint programming techniques within data mining and machine learning. On the other hand, data mining and machine learning could potentially also be used to help constraint programming. Even though there exist some approaches in this direction, we are still far away from a standard use of machine learning and data mining in constraint programming. This tutorial will introduce machine learning and data mining to the constraint programming community. It will provide an overview of several data mining and machine learning tasks, including pattern mining, clustering and probabilistic modeling, and how constraints are used in these tasks, illustrated with the implementation of a number of itemset mining tasks in constraint programming. It will show how data mining and machine learning pose new challenges and opportunities for constraint programming and will address (to a lesser extent) what machine learning and data mining could do for constraint programming.