Nearest prototype classification of noisy data

摘要: Nearest prototype approaches offer a common way to design classifiers. However, when data is noisy, the success of this sort classifiers depends on some parameters that designer needs tune, as number prototypes. In work, we have made study ENPC technique, based nearest approach, in noisy datasets. Previous experimentation algorithm had shown it does not require any parameter tuning obtain good solutions problems where class limits are well defined, and noisy. show able with high classification even A comparison optimal (hand made) other different algorithms demonstrates performance accuracy prototypes noise level increases. We performed experiments four datasets, each them characteristics.