摘要
从异或问题引出泛函联接网络，然后研究泛函联接网络的构造及其数学本质，指出所谓泛函联接均可由神经元的激发函数来实现．最后，关于学习算法提出一点注记，证明该系列论文第2部分提出的学习算法具有一般性或通用性．
The functional-link net is focused. After the XOR problem is introduced,functional-link nets are discussed. Then the structure and the mathematical essence of functional -link nets are studied. It is pointed out that functional-link can be accomplished by means of the activation functions of neurons. At last, on learning algorithms of neural networks, a note is given in which it proved, that the learning algorithms proposed in the second of this series of papers are of generality.
出处
《北京师范大学学报（自然科学版）》
CAS
CSCD
北大核心
1997年第3期305-311,共7页
Journal of Beijing Normal University(Natural Science)
基金
国家自然科学基金!69474014
北京师范大学跨世纪优秀学科带头人培养基金
关键词
神经网络
数学神经网络
泛函联接网络
学习算法
neural networks
mathematical neural networks
functional－link nets
learning algorithms
