摘要
利用粗糙集从数据库中发现规则的能力 ,确定模糊神经网络的初始拓扑结构 ,再利用原始数据对网络进行训练。由于这样构造的模糊神经网络一开始就具备了良好的符合数据分布特性的拓扑结构 ,因而可以大大减小网络规模 ,提高网络训练速度。
A new method of system identification based on rough sets is presented. First, rules are extracted from database using rough set to initialize the structure of neural fuzzy network, then the neural fuzzy network is trained with original data in database. As the advantages of rough sets and neural fuzzy network are combined, the result is very satisfactory. Simulation experiments prove its efficiency.
出处
《华东船舶工业学院学报》
2000年第3期1-6,共6页
Journal of East China Shipbuilding Institute(Natural Science Edition)
基金
国家自然科学基金资助项目! ( 6 984 30 0 2 )
