Adaptation in natural and artificial systems
Adaptation in natural and artificial systems
No free lunch theorems for optimization
IEEE Transactions on Evolutionary Computation
| Hi-index | 0.00 |
A new class of 'crossover' algorithms is proposed that isused in what we coin the Genetic Engineering strategy. These algorithms explicitly consider the fitness of all subsets of candidate solutions when creating the next iteration of candidate solutions. If the fitness of individual solutions is positively correlated to the fitness of their subsets, one could optimize faster by decreasing the probability that good subsets are destroyed while performing recombination. Finding promising partial solutions turns out to be simply a matter of counting. Our implementation, Markov recombination, creates a histogram of all symbol transitions in all candidate solutions in the population at a certain time step. Randomized Markov chains is then be used to generate offspring.