Combining independent and unbiased classifiers using weighted average

• Type: Article in International Conference Proceedings Book
• Date: 2000

Authors

Luís Alexandre (Author) Other; Aurélio Campilho (Author) FEUP; Mohamed Kamel (Author) Other

Conference proceedings International

Title: 15TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION, VOL 2, PROCEEDINGS: PATTERN RECOGNITION AND NEURAL NETWORKS

Pages: 495-498

15th International Conference on Pattern Recognition (ICPR-2000)

BARCELONA, SPAIN, SEP 03-07, 2000

Indexing

ISI Web of Knowledge - 5 Citations

Scopus - 22 Citations

Other information

Authenticus ID: P-001-2BC

DOI: 10.1109/ICPR.2000.906120

Abstract (EN): In a classification problem, improved accuracy can be obtained in many situations by using the combination of several classifiers instead of a single one. In [10], the error reduction that can be obtained by combining unbiased classifiers with independent errors using a simple average, was derived. We present an extension of this result by finding the improvement obtained when combining classifiers using weighted average. We also prove that for unbiased classifiers with independent errors the best combination of N classifiers corresponds to a weighted average, where the combination coefficient of each classifier is equal to 1/N. This means that in these cases the simple average should be used. We present experiments illustrating our results.

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 4