Modular Neural Network Classifiers: A Comparative Study

Authors:
Gasser Auda;Mohamed Kamel
Affiliations:
Pattern Analysis and Machine Intelligence Lab., Systems Design Engineering Department, University of Waterloo, Canada N2L 3G1;Pattern Analysis and Machine Intelligence Lab., Systems Design Engineering Department, University of Waterloo, Canada N2L 3G1
Venue:
Journal of Intelligent and Robotic Systems
Year:
1998

Citing 4
Cited 8

Modular construction of time-delay neural networks for speech recognition

Neural Computation
Democracy in neural nets: voting schemes for classification

Neural Networks
Task decomposition through competition in a modular connectionist architecture

Task decomposition through competition in a modular connectionist architecture
Cooperative modular neural network classifiers

Cooperative modular neural network classifiers

Feature Based Decision Fusion

ICAPR '01 Proceedings of the Second International Conference on Advances in Pattern Recognition
Types of Multinet System

MCS '02 Proceedings of the Third International Workshop on Multiple Classifier Systems
Adaptive fusion and co-operative training for classifier ensembles

Pattern Recognition
Modular neural networks with applications to pattern profiling problems

Neurocomputing
A Hopfield Neural Network for combining classifiers applied to textured images

Neural Networks
Data dependence in combining classifiers

MCS'03 Proceedings of the 4th international conference on Multiple classifier systems
Using an intelligent agent to classify competitor behavior and develop an effective E-market counterstrategy

Expert Systems with Applications: An International Journal
One-against-all ensemble for multiclass pattern classification

Applied Soft Computing

Quantified Score

Hi-index	0.00

Visualization

Abstract

There is a wide variety of Modular Neural Network (MNN) classifiers inthe literature. They differ according to the design of their architecture,task-decomposition scheme, learning procedure, and multi-moduledecision-making strategy. Meanwhile, there is a lack of comparative studiesin the MNN literature. This paper compares ten MNN classifiers which give agood representation of design varieties, viz., Decoupled; Other-output;ART-BP; Hierarchical; Multiple-experts; Ensemble (majority vote); Ensemble(average vote); Merge-glue; Hierarchical Competitive Neural Net; andCooperative Modular Neural Net. Two benchmark applications of differentdegree and nature of complexity are used for performance comparison, and thestrength-points and drawbacks of the different networks are outlined. Theaim is to help a potential user to choose an appropriate model according tothe application in hand and the available computational resources.