Performance Evaluation of an ANN FF Classifier of Raw EEG Data using ROC Analysis
Miguel Antônio Sovierzoski, Fernando Mendes de Azevedo, Fernanda Isabel Marques Argoud · 2008
Due to the increasing use of intelligent and automated systems, pattern, event or signal classification is becoming more important, representing a research area under expansion. The classifier systems indicate results comparable to human classification, or human intervention, with substantial reduction of time and resources. This study presents a methodology for evaluation of an ANN performance using the ROC curve. The statistical performance indexes and the ROC curve are obtained during the supervised training of the ANN FF classifier. The methodology presented was used in the performance evaluation of an ANN classifier of epileptiform events in raw EEG data.