A Case-Based Methodology for Feature Weighting Algorithm Recommendation
Héctor Núòez, Miquel Sànchez–Marrè · Conference on Artificial Intelligence Research and Development · 2005
The underlying idea of the methodology proposed in this paper is to provide a new methodology that could select the most appropriate feature weighting algorithm for a given database. The main idea is to implement a case-based system, where cases are formed by description of databases and feature weighting techniques optimising their generalisation accuracy. This methodology recommends the best techniques to be used in a new unknown database or domain.