A Comparison of Methods for Learning of Highly Non-separable Problems

Authors:
Marek Grochowski;Włodzisław Duch
Affiliations:
Department of Informatics, Nicolaus Copernicus University, Toruń, Poland;Department of Informatics, Nicolaus Copernicus University, Toruń, Poland
Venue:
ICAISC '08 Proceedings of the 9th international conference on Artificial Intelligence and Soft Computing
Year:
2006

Citing 11
Cited 1

Original Contribution: Fast generating algorithm for a general three-layer perceptron

Neural Networks
The target switch algorithm: a constructive learning procedure for feed-forward neural networks

Neural Computation
Learning from Data: Concepts, Theory, and Methods

Learning from Data: Concepts, Theory, and Methods
Binary Rule Generation via Hamming Clustering

IEEE Transactions on Knowledge and Data Engineering
Improvements and Extensions to the Constructive Algorithm CARVE

ICANN 96 Proceedings of the 1996 International Conference on Artificial Neural Networks
A system for induction of oblique decision trees

Journal of Artificial Intelligence Research
Learning highly non-separable Boolean functions using constructive feedforward neural network

ICANN'07 Proceedings of the 17th international conference on Artificial neural networks
Support vector neural training

ICANN'05 Proceedings of the 15th international conference on Artificial neural networks: formal models and their applications - Volume Part II
K-Separability

ICANN'06 Proceedings of the 16th international conference on Artificial Neural Networks - Volume Part I
On sequential construction of binary neural networks

IEEE Transactions on Neural Networks
A constructive algorithm for binary neural networks: the oil-spot algorithm

IEEE Transactions on Neural Networks

Constrained Learning Vector Quantization or Relaxed k-Separability

ICANN '09 Proceedings of the 19th International Conference on Artificial Neural Networks: Part I

Quantified Score

Hi-index	0.00

Visualization

Abstract

Learning in cases that are almost linearly separable is easy, but for highly non-separable problems all standard machine learning methods fail. Many strategies to build adaptive systems are based on the "divide-and-conquer" principle. Constructive neural network architectures with novel training methods allow to overcome some drawbacks of standard backpropagation MLP networks. They are able to handle complex multidimensional problems in reasonable time, creating models with small number of neurons. In this paper a comparison of our new constructive c3sepalgorithm based on k-separability idea with several sequential constructive learning methods is reported. Tests have been performed on parity function, 3 artificial Monks problems, and a few benchmark problems. Simple and accurate solutions have been discovered using c3sepalgorithm even in highly non-separable cases.