Title :
A Dempster-Shafer theory of evidence approach for combining trained neural networks
Author :
Al-Ani, Ahmed ; Deriche, Mohamed
Author_Institution :
Signal Process. Res. Centre, Queensland Univ. of Technol., Brisbane, Qld., Australia
Abstract :
The Dempster-Shafer theory of evidence is a powerful method for combining measures of evidence from different classifiers. However, since there is not a unique way to perform such a combination, we have developed an algorithm which adapts to the training data set so that the overall mean square error is minimised. The proposed method was proved to be superior and more robust than other available combination methods
Keywords :
learning (artificial intelligence); neural nets; pattern classification; Dempster-Shafer theory of evidence; classifiers; combination method; mean square error minimisation; neural network combining; trained neural networks; training data set; Artificial neural networks; Atomic measurements; Australia; Mean square error methods; Neural networks; Pattern recognition; Robustness; Signal processing; Signal processing algorithms; Training data;
Conference_Titel :
Circuits and Systems, 2001. ISCAS 2001. The 2001 IEEE International Symposium on
Conference_Location :
Sydney, NSW
Print_ISBN :
0-7803-6685-9
DOI :
10.1109/ISCAS.2001.921429
