Bounds on the complexity of recurrent ne
โ
Bill G. Horne; Don R. Hush
๐
Article
๐
1996
๐
Elsevier Science
๐
English
โ 823 KB
In this paper the efficiency of recurrent neural network implementations of m-state finite state machines will be explored. Specifically, it will be shown that the node complexity for the unrestricted case can be bounded above by O(v/-m-). It will also be shown that the node complexity is 0( ~) when