Decorate Ensemble of Artificial Neural Networks with High Diversity for Classification?
Journal: International Journal of Computer Science and Mobile Computing - IJCSMC (Vol.2, No. 5)Publication Date: 2013-05-15
Authors : Mittal C. Patel Mahesh Panchal Himani P. Bhavsar;
Page : 134-138
Keywords : Artificial neural network; Classification; Classifier; DECORATE Ensemble; Diversity; Neural Network Ensembles; UCI Datasets;
Abstract
An important data mining task is classification which is used to predict target categories of data instances. DECORATE is one of the most popular ensemble learning techniques, and can use strong learner to build diverse committees in a straightforward strategy. Artificial Neural networks (ANN) are very flexible with respect to incomplete, missing and noisy data and also makes the data to use for dynamic environment. ANN is dependent on how best is the configuration of the net in terms of number of weights, neurons and layers. In this paper, DECORATE with ANN as a base classifier is used to classify data from UCI repository. An experiment is conducted on the public datasets, and the analysis results show that the DECORATE ensemble of ANN improves the performance of classification obviously.
Other Latest Articles
- Multi-Operating Environment System
- SCOPE AND LIMITATION OF ELECTRONIC VOTING SYSTEM
- A Review: Image Extraction with Weighted Page Rank using Partial Tree Alignment Algorithm
- Book Review: Iztok Devetak, Providing High Quality Science Knowledge Using Submicrorepresentations [Zagotavljanje kakovostnega znanja naravoslovja s pomočjo submikroreprezentacij]
- Contemporary Art and Citizenship Education: The Possibilities of Cross-Curricular Links on the Level of Content
Last modified: 2013-05-09 22:30:16