Solving Even-12, -13, -15, -17, -20 and -22 Boolean
Parity Problems using Sub-machine Code GP with Smooth
Uniform Crossover, Smooth Point Mutation and Demes
CSRP-99-2. University of Birmingham, School of Computer Science, (January 1999)Abstract
In this paper we describe a recipe to solve very large
parity problems, using GP without automatically defined
functions. The recipe includes three main ingredients:
smooth uniform crossover (a crossover operator inspired
by theoretical research), sub-machine-code GP (a
technique which allows speeding up fitness evaluation
in Boolean classification problems by nearly 2 orders
of magnitude), and distributed demes (weakly
interacting sub-populations running on separate
workstations). We tested this recipe on parity problems
with up to 22 input variables (i.e. where the fitness
function includes 2^22=4,194,304 fitness cases),
solving them with a very high success probability.