Multiple-prototype classifier design
- 1 February 1998
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man and Cybernetics, Part C (Applications and Reviews)
- Vol. 28 (1), 67-79
- https://doi.org/10.1109/5326.661091
Abstract
Five methods that generate multiple prototypes from labeled data are reviewed. Then we introduce a new sixth approach, which is a modification of Chang's (1974) method. We compare the six methods with two standard classifier designs: the 1-nearest prototype (1-np) and 1-nearest neighbor (1-nn) rules. The standard of comparison is the resubstitution error rate; the data used are the Iris data. Our modified Chang's method produces the best consistent (zero-error) design. One of the competitive learning models produces the best minimal prototypes design (five prototypes that yield three resubstitution errors).Keywords
This publication has 12 references indexed in Scilit:
- Multipactor breakdown in waveguide irisesPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2009
- Nearest prototype classification: clustering, genetic algorithms, or random search?IEEE Transactions on Systems, Man and Cybernetics, Part C (Applications and Reviews), 1998
- Repairs to GLVQ: a new family of competitive learning schemesIEEE Transactions on Neural Networks, 1996
- Minimal consistent set (MCS) identification for optimal nearest neighbor decision systems designIEEE Transactions on Systems, Man, and Cybernetics, 1994
- Initializing the EM algorithm for use in Gaussian mixture modellingPublished by Elsevier BV ,1994
- A possibilistic approach to clusteringIEEE Transactions on Fuzzy Systems, 1993
- Adaptive quadratic neural netsPattern Recognition Letters, 1992
- Competitive learning and soft competition for vector quantizer designIEEE Transactions on Signal Processing, 1992
- Vector Quantization and Signal CompressionPublished by Springer Science and Business Media LLC ,1992
- Self-Organization and Associative MemoryPublished by Springer Science and Business Media LLC ,1989