Use of unsupervised neural networks for classification tasks in electromyography
Authors:
- 1 October 1992
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 3, 1014-1015
Abstract
The present study aims at showing some interesting characteristics of topological feature maps applied to classification tasks in electromyography. Based on Kohonen's model, these self-organizing neural networks are used for the diagnosis of neuromuscular disorders. An in-depth study related to the interpretation of a particular nerve segment has been carried out. Various values of the network parameters such as the number of neurons, number of learning iterations, gain term and neighbor parameter, have been tested, and a comparative study is reported. The advantages of neural network classification are cited.Keywords
This publication has 3 references indexed in Scilit:
- Heterogeneous knowledge representation using a finite automaton and first order logic: a case study in electromyographyArtificial Intelligence in Medicine, 1991
- Self-organized formation of topologically correct feature mapsBiological Cybernetics, 1982
- A Coefficient of Agreement for Nominal ScalesEducational and Psychological Measurement, 1960