146 W.P. Worzel
Fig. 5 Modification of particulate genes from meta-function f
Fig. 6 GPRL with particulate genes
Similarly a function that selects and combines genes based on the values returned
could be implemented. In this scenario, the example in Fig.5 is extended to the
design shown in Fig.6.
3.5 Assembling EvE
Putting all these elements together by using Fog Lifter to provide a sense of place for
the functions in a GP population, adapting ERL to work within GP using an SKGP-
like system and a meta-function that selects and combines particulate genes provides
the basis for a continuous learning system. For example, over time, an agricultural
system could continuously refine crop management recommendations as it moves
from passively learning to actively recommending actions, or a city infrastructure
could adapt to new construction, changing economics, or even macro changes such
as climate-change related issues.