Adaptive Ho-Kashyap rules for perceptron training
- 1 January 1992
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks
- Vol. 3 (1) , 51-61
- https://doi.org/10.1109/72.105417
Abstract
Three adaptive versions of the Ho-Kashyap perceptron training algorithm are derived based on gradient descent strategies. These adaptive Ho-Kashyap (AHK) training rules are comparable in their complexity to the LMS and perceptron training rules and are capable of adaptively forming linear discriminant surfaces that guarantee linear separability and of positioning such surfaces for maximal classification robustness. In particular, a derived version called AHK II is capable of adaptively identifying critical input vectors lying close to class boundaries in linearly separable problems. The authors extend this algorithm as AHK III, which adds the capability of fast convergence to linear discriminant surfaces which are good approximations for nonlinearly separable problems. This is achieved by a simple built-in unsupervised strategy which allows for the adaptive grading and discarding of input vectors causing nonseparability. Performance comparisons with LMS and perceptron training are presented.Keywords
This publication has 10 references indexed in Scilit:
- Autoassociative neural memory capacity and dynamicsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1990
- 30 years of adaptive neural networks: perceptron, Madaline, and backpropagationProceedings of the IEEE, 1990
- Adaptive Dynamic Heteroassociative Neural Memories For Pattern ClassificationPublished by SPIE-Intl Soc Optical Eng ,1989
- High Performance Recording Algorithm For Hopfield Model Associative MemoriesOptical Engineering, 1989
- Dynamic heteroassociative neural memoriesNeural Networks, 1989
- Increased rates of convergence through learning rate adaptationNeural Networks, 1988
- An adaptive attentive learning algorithm for single-layer neural networksPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1988
- Pattern Classifiers and Trainable MachinesPublished by Springer Nature ,1981
- An Algorithm for Linear Inequalities and its ApplicationsIEEE Transactions on Electronic Computers, 1965
- A generalized inverse for matricesMathematical Proceedings of the Cambridge Philosophical Society, 1955