Evolving Classifiers: Methods for Incremental Learning - Computer Science > LearningReportar como inadecuado




Evolving Classifiers: Methods for Incremental Learning - Computer Science > Learning - Descarga este documento en PDF. Documentación en PDF para descargar gratis. Disponible también para leer online.

Abstract: The ability of a classifier to take on new information and classes byevolving the classifier without it having to be fully retrained is known asincremental learning. Incremental learning has been successfully applied tomany classification problems, where the data is changing and is not allavailable at once. In this paper there is a comparison between Learn++, whichis one of the most recent incremental learning algorithms, and the new proposedmethod of Incremental Learning Using Genetic Algorithm ILUGA. Learn++ hasshown good incremental learning capabilities on benchmark datasets on which thenew ILUGA method has been tested. ILUGA has also shown good incrementallearning ability using only a few classifiers and does not suffer fromcatastrophic forgetting. The results obtained for ILUGA on the OpticalCharacter Recognition OCR and Wine datasets are good, with an overallaccuracy of 93% and 94% respectively showing a 4% improvement over Learn++.MTfor the difficult multi-class OCR dataset.



Autor: Greg Hulley, Tshilidzi Marwala

Fuente: https://arxiv.org/







Documentos relacionados