Analysis on the learning ability of improved sparse distributed memory model

被引:0
|
作者
Peng, Hong-Jing [1 ]
Chen, Song-Can [1 ]
机构
[1] Dept. of Comp. Sci. and Eng., Nanjing Univ. of Aero. and Astron., Nanjing 210016, China
来源
关键词
Feedforward neural networks - Learning systems - Mathematical models;
D O I
暂无
中图分类号
学科分类号
摘要
Kanerva's sparse distributed memory model (SDM) is initially applied to associative memory, its application is limited for its reading-writing operation by using outer product. SDM is improved by changing its original rule of reading-writing operation and retaining its characteristics of sparse distributed memory, then a novel improved model that is similar to CMAC will be obtained, in which no block effect appear and Hashing technology is not used. Theoretical analysis and example show that this improved model is effective and reasonable.
引用
收藏
页码:774 / 776
相关论文
共 50 条