rEDA: reverse estimation of distribution algorithm for classification
by Jian-cong Fan; Zheng Feng; Wen-hua Liu; Yu-hao Cai; Yong-quan Liang
International Journal of Innovative Computing and Applications (IJICA), Vol. 6, No. 3/4, 2015

Abstract: Estimation of distribution algorithm (EDA) is a branch of evolutionary algorithms. EDA replaces recombination and mutation operators with the estimation of probabilistic distribution of selected individuals. However, these selected individuals only cover part of the problem to be optimised, which causes that the algorithm may easily fall into a local optimum. In this paper, we propose a variation of EDA, reverse estimation of distribution algorithm (rEDA), from the perspective of reverse process. Different from the EDA process that individuals are firstly given and then the estimation of models starts, rEDA is to firstly give initial models and then regulate these models relying on sampling from the models and optimisation objective. We employ rEDA to classification in data mining area and propose a novel classification algorithm based on rEDA. The proposed rEDA algorithm and rEDA-based classification algorithm are analysed theoretically. The empirical results show our proposed algorithm outperforms some classical classification algorithms in accuracy.

Online publication date: Wed, 11-Nov-2015

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Innovative Computing and Applications (IJICA):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com