Pattern classification by stochastic neural network with missing data

Author

Tanaka, Masahiro ; Kotokawa, Yasuaki ; Tanino, Tetsuzo

Author_Institution

Dept. of Inf. Technol., Okayama Univ., Japan

Volume

1

fYear

1996

fDate

14-17 Oct 1996

Firstpage

690

Abstract

In this paper, pattern classification by stochastic neural networks is considered. This model is also called a Gaussian mixture model. When missing data exist in the training data, it is usual to remove incomplete instants. Here we take another approach, where the missing elements are estimated by using the conditional expectation based on the estimated model by using the EM algorithm. It is shown by using Fisher´s Iris data that this approach is superior to removing incomplete data

Keywords

learning (artificial intelligence); neural nets; parameter estimation; pattern classification; probability; EM algorithm; Fisher´s Iris data; Gaussian mixture model; conditional expectation; expectation maximisation; missing data; pattern classification; stochastic neural network; Function approximation; Information technology; Iris; Multi-layer neural network; Neural networks; Parameter estimation; Pattern classification; Probability density function; Stochastic processes; Training data;

fLanguage

English

Publisher

ieee

Conference_Titel

Systems, Man, and Cybernetics, 1996., IEEE International Conference on

Conference_Location

Beijing

ISSN

1062-922X

Print_ISBN

0-7803-3280-6

Type

conf

DOI

10.1109/ICSMC.1996.569878

Filename

569878