【摘要】 训练模式对的小幅摄动可能对模糊神经网络的性能产生副作用,为此文中提出了一般性的模糊神经网络对训练模式对摄动的鲁棒性概念,并就典型的模糊双向联想记忆网络FBAM进行了具体分析.理论研究表明FBAM采用模糊赫布学习算法时该鲁棒性好,而采用新近提出的另一学习算法时,该鲁棒性较差.为此,作者为后一算法提供了一种训练模式对摄动的控制方法,以保证FBAM的这种鲁棒性较好.最后用FBAM在图像联想方面的实验证实了文中的某些理论结果.文中工作对FBAM系统的性能分析、学习算法的选择和模式对获取过程的指导有一定意义.更多还原

【Abstract】 Small perturbations of training pattern pairs may cause some disadvantage to performance of a fuzzy neural network (FNN), so a new concept is established for the robustness of an FNN to perturbations of training pattern pairs. As a typical instance, this kind of robustness of fuzzy bidirectional associative memory (FBAM) is analyzed, and the theoretical studies in this paper show that the FBAM using the Hebbian learning algorithm has good such robustness, however, the FBAM using another learning algorithm presented recently has poor such robustness. A method to control perturbations of pattern pairs is proposed in order to make such robustness of FBAM be good when the latter learning algorithm is employed. Finally, some theoretical results in the paper are also confirmed by the given experiment in which the FBAM associates an image with another image. The work in the paper is of some benefit to the performance analyses, the choice of learning algorithms, and the guidance to pattern pair acquisition for FBAM.更多还原