Multimedia & Applications

Biography

Biography: Dimitrios A. Karras

Abstract

A novel methodology is herein outlined for multimedia data mining problems by designing an hierarchical pattern mining neural system. The proposed system combines the data mining decisions of different neural network pattern mining systems. Instead of the usual approach for applying voting schemes on the decisions of their output layer neurons, the proposed methodology integrates higher order features extracted by their upper hidden layer units. More specifically, different instances (cases) of each such pattern mining system, derived from the same training process but with different training parameters, are investigated in terms of their higher order features, through similarity analysis, in order to find out repeated and stable higher order features. Then, all such higher order features are integrated through a second stage neural network pattern mining system having as inputs suitable similarity features of them. The herein suggested hierarchical pattern mining neural system for multimedia data mining applications shows improved pattern mining performance in series of experiments in computer vision databases and face recognition databases . The validity of this novel combination approach of pattern mining neural systems has been investigated when the first stage neural pattern mining systems involved correspond to different Feature Extraction Methodologies (FEM) for either shape or face classification. The experimental study illustrates that such an approach, integrating higher order features through similarity analysis of a committee of the same pattern mining instances (cases) and a second stage neural pattern mining integration system, outperforms other combination methods, like voting combination schemes as well as single neural network pattern mining systems having as inputs all FEMs derived features. In addition, it outperforms hierarchical combination methods non performing integration of cases through similarity analysis