Is the common good? a new perspective developed in genetic algorithms
Stephen Yuangi Chen, Stephen F. Smith · 1999
. The Commonality-Based Crossover Framework redefines crossover as a two step process: 1) preserve the maximal common schema of two parents, and 2) complete the solution with a construction heuristic. To demonstrate the utility of this design model, domainindependent operators, heuristic operators, and hybrid operators have been developed for benchmark and practical problems with standard and non-standard representations. The new commonality-based operators have performed consistently better than comparable operators which emphasize combination. In heuristic operators (which use problem specific heuristics during crossover), the effects of commonality-based selection have been isolated in GENIE (a genetic algorithm that eliminates fitness-based selection of parents). Since the effectiveness of construction heuristics can be amplified by using only commonality-based restarts, the preservation of common components has supplied selective pressure at the component (rather than individual