FITNESS FUNCTION
| Collection | International Encyclopedia of Systems and Cybernetics |
|---|---|
| Year | 2004 |
| Vol. (num.) | 2(1) |
| ID | ◀ 1282 ▶ |
| Object type | Discipline oriented, Methodology or model |
The criteria by which sets ofrules can be selected to create an evolutionary process aimed at evolving a self-replicating automaton .
M. SIPPER and J.A. REGGIA (2001) devised a “fitness function composed of a weighted sum of three measures: a growth measure (the extend to which each component type generates an increasing supply of that component), a relative position measure (the extend to which neighboring components stay together) and a replicant measure (a function of the number of the actual replicatory elements present)
With the right fitness function, evolution can turn rule sets that are sterile into ones that are fecund; the process usually takes 150 or so generations“.
- “The ability to replicate without a self-description may be relevant to questions about how the earliest biological replicators originated. In a sense, researchers are seeing a continuum between non-living and living structures ”(p. 31-35)
See also
Artificial life, Cybernetics (2$^{nd}$ Evolution (Artificial), Game of life (Conway), Neural network, Paralled distributed processing, Weights (Synaptic)