Title :
Two variations on Fisher´s linear discriminant for pattern recognition
Author_Institution :
Center for Sensor Signal & Inf. Process., Mawson Lakes, SA, Australia
fDate :
2/1/2002 12:00:00 AM
Abstract :
Discriminants are often used in pattern recognition to separate clusters of points in some multidimensional "feature" space. The paper provides two fast and simple techniques for improving on the classification performance provided by Fisher\´s linear discriminant for two classes. Both of these methods are also extended to nonlinear decision surfaces through the use of Mercer kernels
Keywords :
decision theory; learning automata; pattern classification; probability; search problems; Fisher linear discriminant; Mercer kernels; classification performance; clusters; learning automata; multidimensional feature space; nonlinear decision surfaces; pattern recognition; support vector machines; Fractals; Histograms; Kernel; Linear discriminant analysis; Multidimensional systems; Pattern recognition; Radar detection; Robustness; Testing; Vehicle detection;
Journal_Title :
Pattern Analysis and Machine Intelligence, IEEE Transactions on
