Multi-class minimax probability machine

This paper investigates the multi-class Minimax Probability Machine (MPM). MPM constructs a binary classifier that provides a worst-case bound on the probability of misclassification of future data points, based on reliable estimates of means and covariance matrices of the classes from the training data points. We propose a method to adapt MPM to multi-class datasets using the one-against-all strategy. And then we introduce an optimal kernel for MPM for each specific dataset found by Genetic Algorithms (GA) [1]. The proposed method was evaluated on stomach cancer data. The obtained results are better and more stable than for using a single kernel.

Title: 

Multi-class minimax probability machine
Authors: Dang, Tat-Dat; Nguyen, Ha-Nam
Keywords: Genetic algorithms; Minimax probability machine; One-against-all; One-against-one
Issue Date: 2009
Publisher: H:Đại học Quốc gia Hà Nội
Abstract: This paper investigates the multi-class Minimax Probability Machine (MPM). MPM constructs a binary classifier that provides a worst-case bound on the probability of misclassification of future data points, based on reliable estimates of means and covariance matrices of the classes from the training data points. We propose a method to adapt MPM to multi-class datasets using the one-against-all strategy. And then we introduce an optimal kernel for MPM for each specific dataset found by Genetic Algorithms (GA) [1]. The proposed method was evaluated on stomach cancer data. The obtained results are better and more stable than for using a single kernel. © 2009 IEEE.
Description: 1st International Conference on Knowledge and Systems Engineering, KSE 2009; Hanoi; Viet Nam; 13 October 2009 through 17 October 2009; Category numberE3846; Code 79895
URI: http://repository.vnu.edu.vn/handle/VNU_123/32091
ISSN: 978-076953846-4
Appears in Collections:Bài báo của ĐHQGHN trong Scopus

Nhận xét

Bài đăng phổ biến từ blog này

Development and regulation of pedicel abscission in tomato

Study of eta-eta ' mixing from measurement of B-(s)(0) -> J/psi eta((')) decay rates