×

Detecting interesting instances. (English) Zbl 1019.68736

Hand, David J. (ed.) et al., Pattern detection and discovery. ESF exploratory workshop, London, UK, September 16-19, 2002. Proceedings. Berlin: Springer. Lect. Notes Comput. Sci. 2447, 13-23 (2002).
Summary: Most valid rules that are learned from very large and high dimensional data sets are not interesting, but are already known to the users. The dominant model of the overall data set may well suppress the interesting local patterns. The search for interesting local patterns can be implemented by a two step learning approach which first acquires the global models before it focuses on the rest in order to detect local patterns. In this paper, three sets of interesting instances are distinguished. For these sets, the hypothesis space is enlarged in order to characterize local patterns in a second learning step.
For the entire collection see [Zbl 1007.68737].

MSC:

68U99 Computing methodologies and applications
68P15 Database theory
68P20 Information storage and retrieval of data
68T10 Pattern recognition, speech recognition
68T05 Learning and adaptive systems in artificial intelligence
PDFBibTeX XMLCite
Full Text: Link