A Robust Hierarchical Neural Network Methodology for Improved Image Classification Performance

Abstract

A novel methodology is herein presented for combining the decisions of different feedforward neural network classifiers. Instead of the usual approach of applying voting schemes on the decisions of their output layer neurons, the proposed methodology integrates the higher order features extracted by their upper hidden layer units through a second stage feedforward neural network having as inputs all such higher order features. Therefore, an hierarchical neural system for pattern recognition has been developed with improved classification performance. The validity of this novel combination approach has been investigated when the first stage neural classifiers involved correspond to different Feature Extraction Methodologies (FEM) for shape classification. The experimental study illustrates that such an approach, integrating higher order features into a second stage feedforward neural classifier, outperforms other combination methods, like voting combination schemes as well as single neural network classifiers having as inputs all FEMs derived features.