Journal
EXPERT SYSTEMS WITH APPLICATIONS
Volume 37, Issue 10, Pages 6904-6918Publisher
PERGAMON-ELSEVIER SCIENCE LTD
DOI: 10.1016/j.eswa.2010.03.031
Keywords
Cooperative coevolutionary algorithms; RBFNN; Feature selection; Multiclass classification
Categories
Funding
- Research Fund for the Doctoral Program of Higher Education of China [20090032110065, 20090032120073]
- National Science Fund for Distinguished Young Scholars of China [70925005]
Ask authors/readers for more resources
There are irrelevant features that are redundant or significantly degrade the learning accuracy in the real-world complex classification tasks. This paper presents a new hybrid learning algorithm based on a cooperative coevolutionary algorithm (Co-CEA) with dual populations for designing the radial basis function neural network (RBFNN) models with an explicit feature selection. This approach attempts to complete both the RBFNN construction and the feature selection simultaneously. The proposed algorithm utilizes the Co-CEA's divide-and-cooperative mechanism, which utilizes the evolutionary algorithms executing in parallel to coevolve subpopulations, corresponding to the hidden layer structure and the dominate features respectively. The algorithm adopts the binary encoding to represent the feature subset and the matrix-form mixed encoding to represent the RBFNN hidden layer structure, and a complete solution is formed via collaborations among the two subpopulations. Experimental results illustrate that the proposed algorithm outperforms other algorithms in references in terms of the classification accuracy, and it is able to obtain both prominent features and good RBFNN structure with higher prediction capability. (C) 2010 Elsevier Ltd. All rights reserved.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available