DYNAMIC SUBSET SELECTION APPLIED TO SELF-ORGANIZING MAPS

摘要: The self-organizing map (SOM) is an unsupervised learning algorithm that attempts to form a compact representation of data set via prototype vectors exist in the same space. Dynamic subset selection (DSS) genetic programming (GP) based method selecting particularly difficult-to-learn patterns from set, where difficulty GP-specific measure. In this work, dynamic (DSSSOM) presented. It application DSS SOM, with modifications both. able handle very large sets, and has DSS-based built-in stopping mechanism. performance new measured over five sets original implementation, compared SOM other algorithms. Results show DSSSOM achieves on par training time reduced by factor up nearly hundred.