Title :
A new algorithm for neural network architecture study
Author :
Qin, Zhongguang ; Mao, Zongyuan
Author_Institution :
Dept. of Autom., South China Univ. of Technol., Guangzhou, China
Abstract :
The neural network architecture study is one of the most difficult tasks. This paper presents a new algorithm for feedforward neural network architecture using a systematic approach. This approach is based on rough set and knowledge representation of neural network. This algorithm can fix the optimal number of hidden layer units. This new algorithm is also tested in Chinese medicine practical projects. The results show that the new algorithm is of wide applicability for the study of neural network architecture
Keywords :
feedforward neural nets; knowledge representation; medical diagnostic computing; medical expert systems; neural net architecture; optimisation; rough set theory; Chinese medicine practical projects; feedforward neural network architecture; hidden layer units; knowledge representation; rough set; Artificial intelligence; Artificial neural networks; Automation; Computer networks; Feedforward neural networks; Knowledge representation; Mathematics; Medical tests; Neural networks; Testing;
Conference_Titel :
Intelligent Control and Automation, 2000. Proceedings of the 3rd World Congress on
Conference_Location :
Hefei
Print_ISBN :
0-7803-5995-X
DOI :
10.1109/WCICA.2000.863338
