A Modular Neural Network Architecture with Additional Generalization Abilities for Large Input Vectors

Albrecht Schmidt
The Intelligent Systems Group
Department of Computing
The Manchester Metropolitan University

Zuhair Bandar
The Intelligent Systems Group
Department of Computing
The Manchester Metropolitan University

Abstract:

This paper proposes a two layer modular neural system. The basic building blocks of the architecture are multilayer Perceptrons trained with the Backpropagation algorithm.

Due to the proposed modular architecture the number of weight connections is less than in a fully connected multilayer Perceptron.

The modular network is designed to combine two different approaches of generalization known from connectionist and logical neural networks; this enhances the generalization abilities of the network.

The architecture introduced here is especially useful in solving problems with a large number of input attributes.

 
• Introduction
• The Network Architecture
• Training the System
• Calculation of the Output
• On Generalization
• Experiments
• Limitations
• Conclusion
• References
• About this document ...