×

Concept learning with approximation: Rough version spaces. (English) Zbl 1013.68574

Alpigini, James J. (ed.) et al., Rough sets and current trends in computing. 3rd international conference, RSCTC 2002, Malvern, PA, USA, October 14-16, 2002. Proceedings. Berlin: Springer. Lect. Notes Comput. Sci. 2475, 239-246 (2002).
Summary: The concept learning problem is a general framework for learning concept consistent with available data. Version Spaces theory and methods are build in this framework. However, it is not designated to handle noisy (possibly inconsistent) data. In this paper, we use rough set theory to improve this framework. Firstly, we introduce a rough consistency. Secondly, we define an approximative concept learning problem. Thirdly, we present a Rough Version Space theory and related methods to address the approximative concept learning problem. Using a didactic example, we put these methods into use. An overview of possible extension of this work concludes this article.
For the entire collection see [Zbl 1001.00048].

MSC:

68T37 Reasoning under uncertainty in the context of artificial intelligence
68T05 Learning and adaptive systems in artificial intelligence
PDFBibTeX XMLCite
Full Text: Link