A Robust Adaptive Version of Evidence-Theoretic k-NN Classification Rule

Authors:
Zhi-gang Su;Pei-hong Wang
Affiliations:
-;-
Venue:
FSKD '09 Proceedings of the 2009 Sixth International Conference on Fuzzy Systems and Knowledge Discovery - Volume 04
Year:
2009

Citing 0
Cited 2

Maximal confidence intervals of the interval-valued belief structure and applications

Information Sciences: an International Journal
Minimizing neighborhood evidential decision error for feature evaluation and selection based on evidence theory

Expert Systems with Applications: An International Journal

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, a robust adaptive version of evidence theoretic k-NN classification rule was proposed. In the robust rule, an adaptive distance metric was proposed to be used instead of the Euclidean distance metric. All the parameters brought in by the proposed adaptive distance metric and some other important structural parameters fixed in the original rule are optimized based on training set by means of gradient-descent algorithm. In addition, a new error criterion and also an extended form of combination rule were proposed to be applied. Some popular sets of data were applied to validate the robust adaptive version of evidence-theoretic rule, and the results suggest that the robust one outperforms the original one.