The application of EAs can be prohibitive when the evaluations
are computationally expensive or an explicit mathematical
fitness function is unavailable. Whilst the speed and
cost of rapid-prototyping continues to improve, fabricating
an evolved design before fitness can be assigned remains an
expensive task when potentially thousands of evaluations are
required (e.g., 10 min print time for each very simple individual
in [50]). However, given a sample D of evaluated
individuals N, a surrogate model, y = f (x), can be constructed,
where x is the genotype, and y, fitness, in order to compute
the fitness of an unseen data point x / ∈ D. The use of surrogate
models has been shown to reduce the convergence time
in evolutionary computation and multiobjective optimization;
see [51]–[53] for recent reviews. Alternative methods, such
as fitness inheritance [54], fitness imitation [55], and fitness
assignment [56] can also be used.
Typically, a set of evaluated genotypes
The application of EAs can be prohibitive when the evaluationsare computationally expensive or an explicit mathematicalfitness function is unavailable. Whilst the speed andcost of rapid-prototyping continues to improve, fabricatingan evolved design before fitness can be assigned remains anexpensive task when potentially thousands of evaluations arerequired (e.g., 10 min print time for each very simple individualin [50]). However, given a sample D of evaluatedindividuals N, a surrogate model, y = f (x), can be constructed,where x is the genotype, and y, fitness, in order to computethe fitness of an unseen data point x / ∈ D. The use of surrogatemodels has been shown to reduce the convergence timein evolutionary computation and multiobjective optimization;see [51]–[53] for recent reviews. Alternative methods, suchas fitness inheritance [54], fitness imitation [55], and fitnessassignment [56] can also be used.Typically, a set of evaluated genotypes
การแปล กรุณารอสักครู่..
