NETWORK (Parallel- processing)
| Collection | International Encyclopedia of Systems and Cybernetics |
|---|---|
| Year | 2004 |
| Vol. (num.) | 2(2) |
| ID | ◀ 2269 ▶ |
| Object type | General information, Methodology or model |
A network in which “all elements compute their next activities at the same moment”.
S. KAUFFMAN comments about this type of networks: “If we find order in random networks, than random parallel networks with random logic have order despite an apparent cacophony of structure and logic” (1993, p.192).
This order would be most probably ergodic.
KAUFFMAN also observes: “… the current enthousiasm for connectionism and for parallel-processing neural networks is precisely that any such system almost inevitably classifies,… has internal dynamics whose attractors represent alternative asymptotic states of the network” (p.233). And the following statement may possibly be considered as another formulation of autopoiesis: “The alternative attractors in a fixed environment from which the network receives inputs can be thought of as alternative classification of the same environment. Similarly, alternative environments which map to the same attractor can be seen as classifications of those different environments as ”the same“ by the network. Thus such networks inevitably classify and have internal models of their worlds” (Ibid).