An Experimental Evaluation of Coevolutive Concept Learning
Abstract
In this paper an extensive experimental evaluation of an evolutionary approach to concept learning is presented. The experimentation, performed with the system G-NET, investigates the effectiveness of the approach along the following dimensions: Robustness with respect to parameter setting, effectiveness of the MDL criterion coupled with a stochastic search bias, impact of coevolution on the quality of the solution and on the computational effort required, and ability to face problems requiring structured representation languages. A discussion of the obtained results and a suggestion on when this type of approach might be useful is also provided. 1 INTRODUCTION Supervised concept learning has been tackled, so far, with several approaches, including symbolic, connectionist and evolutive ones. Different approaches are better suited to different classes of problems, depending, for instance, on the nature of data or the availability of domain-specific knowledge. In the hope of making a li...
Cite
Text
Anglano et al. "An Experimental Evaluation of Coevolutive Concept Learning." International Conference on Machine Learning, 1998.Markdown
[Anglano et al. "An Experimental Evaluation of Coevolutive Concept Learning." International Conference on Machine Learning, 1998.](https://mlanthology.org/icml/1998/anglano1998icml-experimental/)BibTeX
@inproceedings{anglano1998icml-experimental,
title = {{An Experimental Evaluation of Coevolutive Concept Learning}},
author = {Anglano, Cosimo and Giordana, Attilio and Bello, Giuseppe Lo and Saitta, Lorenza},
booktitle = {International Conference on Machine Learning},
year = {1998},
pages = {19-27},
url = {https://mlanthology.org/icml/1998/anglano1998icml-experimental/}
}