SPR_LectureHandouts_QuantifyingClassificationPerformance

SPR_LectureHandouts_QuantifyingClassificationPerformance -...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Pattern Recognition ECE-8443 Quantifying Classification Performance Electrical and Computer Engineering Department, Mississippi State University. 1 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department Outline Confusion Matrices Overall Accuracy, False Alarm Rate, Sensitivity, Specificity Receiver Operating Characteristics 2 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department Confusion Matrices 3 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department Confusion Matrices 4 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department Confusion Matrices 5 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department Potential Problems with Confusion Matrices 6 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department Confusion Matrices – Extending to a general c-class recognition task. class1 class3 class4 class5 class6 class7 % class1 87 0 0 0 0 0 0 100 class2 1 73 15 5 0 0 0 77.7 class3 4 10 67 7 3 1 0 72.8 class4 1 7 12 62 6 1 0 69.7 class5 3 0 2 12 55 2 8 67.1 class6 18 0 1 4 6 42 4 56.0 class7 5 5 1 1 3 0 66 81.5 % 7 class2 73 77 68 68 75 91 84 75.3 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department Receiver Operating Characteristics (ROC) 8 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department ROC Space and ROC Curves: TPR vs. FPR 9 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department ROC Space and ROC Curves: TPR vs. FPR 10 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department ROC Space and ROC Curves: TPR vs. FPR Area under the ROC curve can potentially be a good metric for measuring the usefulness of a feature space for a binary classification task 11 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department Properties of ROC curves 12 Performance Measures Saurabh Prasad Pattern Recognition Electrical and Computer Engineering Department ...
View Full Document

This note was uploaded on 02/20/2012 for the course ECE 8443 taught by Professor Staff during the Spring '10 term at University of Houston.

Ask a homework question - tutors are online