Title :
Applying Static and Dynamic Weight Measures in Ensemble Systems
Author :
Paradeda, Raul ; Xavier, Joao C. ; Canuto, Anne M P
Author_Institution :
Inf. & Appl. Math. Dept., Fed. Univ. of Rio Grande do Norte, Natal
Abstract :
It is well known that the use of ensemble systems usually increases the accuracy rate of individual machine learning systems. A way of improving the accuracy of these systems even further is through the use of weight measures. This paper analyzes the influence of the use of static and dynamic weights in the accuracy of two structures (homogeneous and heterogeneous) of ensemble systems. Furthermore, it investigates the relation between diversity and the use weights in ensemble system.
Keywords :
learning (artificial intelligence); pattern classification; dynamic weight measures; ensemble system; heterogeneous structure; homogeneous structure; machine learning system; multiclassifier system; static weight measures; Classification algorithms; Diversity reception; Informatics; Learning systems; Mathematics; Neural networks; Pattern recognition; Robustness; Testing; Weight measurement; Dynamic Weights; Ensemble Systems;
Conference_Titel :
Neural Networks, 2008. SBRN '08. 10th Brazilian Symposium on
Conference_Location :
Salvador
Print_ISBN :
978-1-4244-3219-6
Electronic_ISBN :
1522-4899
DOI :
10.1109/SBRN.2008.35
