Data classification with multilayer perceptrons using a generalized error function

Authors:
Luís M. Silva;J. Marques de Sá;Luís A. Alexandre
Affiliations:
INEB-Instituto de Engenharia Biomédica, Porto, Portugal;INEB-Instituto de Engenharia Biomédica, Porto, Portugal and Faculdade de Engenharia, Universidade do Porto, Porto, Portugal;Dep. de Informática, Universidade da Beira Interior, Covilhã, Portugal and IT, Networks and Multimedia Group, Covilhã, Portugal
Venue:
Neural Networks
Year:
2008

Citing 6
Cited 3

Accelerated learning in layered neural networks

Complex Systems
Connectionist learning procedures

Artificial Intelligence
New developments of the Z-EDM algorithm

ISDA '06 Proceedings of the Sixth International Conference on Intelligent Systems Design and Applications - Volume 01
Adaptive mixtures of local experts

Neural Computation
Neural network classification: maximizing zero-error density

ICAPR'05 Proceedings of the Third international conference on Advances in Pattern Recognition - Volume Part I
A novel objective function for improved phoneme recognition using time-delay neural networks

IEEE Transactions on Neural Networks

The Influence of the Risk Functional in Data Classification with MLPs

ICANN '08 Proceedings of the 18th international conference on Artificial Neural Networks, Part I
A novel hybrid classification model of artificial neural networks and multiple linear regression models

Expert Systems with Applications: An International Journal
Mean multiclass type i and II errors for training multilayer perceptron with particle swarm in image segmentation

IDEAL'12 Proceedings of the 13th international conference on Intelligent Data Engineering and Automated Learning

Quantified Score

Hi-index	0.00

Visualization

Abstract

The learning process of a multilayer perceptron requires the optimization of an error function E(y,t) comparing the predicted output, y, and the observed target, t. We review some usual error functions, analyze their mathematical properties for data classification purposes, and introduce a new one, E"E"x"p, inspired by the Z-EDM algorithm that we have recently proposed. An important property of E"E"x"p is its ability to emulate the behavior of other error functions by the sole adjustment of a real-valued parameter. In other words, E"E"x"p is a sort of generalized error function embodying complementary features of other functions. The experimental results show that the flexibility of the new, generalized, error function allows one to obtain the best results achievable with the other functions with a performance improvement in some cases.