摘要
通用并行神经网络模拟系统GP^2N^2S^2的算法库中,目前已集成了BP、Kohonen、Hopfield等较为成熟和流行的网络算法,并且已被用来成功地解决了诸如TSP、联想记忆(AM)、镜象对称等实际问题。我们还用动态代码加载的方式实现了用户自定义算法的模拟调用。
There are several popular algorithms of neural network in the algorithmic library of GP2N2S2 such as BP, Kohonen and Hopfield. We have used them to solve some actual tasks such as travelling salesman problem (TSP), Associative Memory (AM) and mirror symmetry. We have also successed in simulation calling of algorithm designed by user with dynamic code loading.
出处
《小型微型计算机系统》
CSCD
北大核心
1992年第12期37-43,48,共8页
Journal of Chinese Computer Systems
基金
国家科委基础研究高技术司
863计划的资助
关键词
神经网络
学习算法
GP^2N^2S^2
Neural networks Learning algorithm Transputer array Dynamic code loading