Auto Target Multiclassification by Double Negative Aggregation of SVM Membership(Remote sensing/Radar (2), Workshop for Space, Aeronautical and Navigational Electronics (WSANE 2005))

概要

論文の詳細を見る
Auto Target Multiclassification problems are binarized into pairwise target classifications to utilize basic classification methods such as the support vector machine (SVM). Instead of the fuzzy logical product, we propose a simple aggregation called double negative. It focuses on the margin areas cf the SVM discrimination functions, and the memberships of the negative votes of the class are accumulated to be the negative membership of the class. It provides consistent results with the basic pairwise memberships, enumerates candidates when the total membership of the multiple classes are nearly equal to each other, and needs small computational cost in class reconfiguration.
社団法人電子情報通信学会の論文
2005-02-26