Information Technology Reference
In-Depth Information
OUT
OUT
IN1
IN2
AND1
AND2
OR1
OR2
IN1
IN2
AND1
AND2
OR1
OR2
OUT
OUT
IN1
I
N2
IN2
IN1
∨
∧
∧
OUT = ( IN1
IN2 )
(IN1
IN2 )
IN1
IN2
AND1
AND2
OR1
OR2
Figure 3.5
Development of a logic circuit from a simple genotype. The output termi-
nal receptor binds an AND1 enzyme as its substrate. The AND1 enzyme now chooses
its own substrates, and the process continues until the inputs of all expressed enzymes
have been satisfied. Note that OR1 is never expressed, and IN1 and IN2 are both
expressed twice.
There are potentially many ways of implementing this process. The im-
plementation used in this study is a simple breadth-first search. Development
begins with expression of the receptors, which then choose substrates whose
shapes are most similar to their specificities. Substrates are chosen from those
defined in the genotype and can be either enzymes or glands. These substrates
are now considered expressed and, if they require inputs, attempt to satisfy them
by binding their own substrates. This process continues in hierarchical fashion
until all expressed receptors and enzymes have satisfied all of their inputs. An
example of the process is illustrated in Figure 3.5 for the development of a
combinational logic circuit. It is interesting to note that development does not
require the executional structure to contain all the components described within
the genome. Also, enzymes that are expressed may be used as input sources for
more than one other enzyme.
Functionality as Shape
Recall that one of the problems with the parse tree representation of conven-
tional GP is that context, which is determined by position, is not preserved by
subtree crossover. In enzyme GP, the position of an enzyme within the genome
