摘要
针对在系统辨识和控制领域中广为应用的Elman网络 (尤其是对角回归网络 )中的学习算法重要问题进行了论述。并就当前文献中存在的三种留传颇广的梯度学习算法进行了讨论 ,指出其中两种算法所存在的问题并对算法的实质进行了初步分析。
Learning Algorithms, the important problems of a local recurrent Elman neural network in the field of system identification and control, are discussed. Three learning algorithms based on gradients are analyzed and compared. Problems existing in two of these algorithms are pointed out and the essence of the algorithms is analyzed.
出处
《武汉理工大学学报(信息与管理工程版)》
CAS
2002年第2期14-15,41,共3页
Journal of Wuhan University of Technology:Information & Management Engineering
基金
国家自然科学基金资助项目 (79990 85 0 )