Classification-assisted memetic algorithms for solving optimization problems with restricted equality constraint function mapping

A Classification-Based Surrogate-Assisted Evolutionary Algorithm for Expensive Many-Objective Optimization

“…During the environmental selection in CRADE, offspring solutions worse than their parents are discarded according to the classification surrogate, and the fitness of the rest offspring solutions are predicted by the regression surrogate. In [34], classification was used to assist a memetic algorithm in choosing individuals to be refined for solving optimization problems with single equality constraint. In this algorithm, a support vector machine was used to determine whether a solution is close to the feasible region and whether local refinement should be carried out.…”

“…Various types of surrogates are commonly used in expensive optimization, including polynomial response surface methodology [27], radial basis function [28], Gaussian process model, also known as Kriging model [29], or sometimes as efficient global optimization (EGO), artificial neural networks [30], and support vector machines [31]. A variety of surrogateassisted evolutionary algorithms (SAEAs) were proposed to handle single-objective optimization using classification or regression based fitness approximation, e.g., the neural network assisted evolution strategy [32], the feasibility structure modeling assisted memetic algorithm [33], the classificationassisted memetic algorithm [34], and the surrogate-assisted cooperative particle swarm optimization [35]. Furthermore, many SAEAs for expensive multi-objective optimization were proposed in the past decades, e.g., the generalized surrogateassisted multi-objective memetic algorithm (GS-MOMA) [36], the weighted aggregation based multi-objective optimization assisted by efficient global optimization (ParEGO) [37], the efficient global optimization assisted MOEA/D (MOEA/D-EGO) [21], the Pareto rank learning MOEA [38], and the Kriging assisted RVEA (K-RVEA) [39], for solving MaOPs.…”

Model-based evolutionary algorithms: a short survey

“…Apart from the aforementioned single-objective SAEAs that work in the context of genetic algorithm, there are also some SAEAs working in stochastic search methods rather than genetic algorithms, such as the surrogate-assisted artificial immune systems [69], the neural network-assisted evolution strategy [57], the feasibility structure modelingassisted memetic algorithm [70], the classification-assisted memetic algorithm [71], the surrogate-assisted cooperative particle swarm optimization [72], and the committee-based active learning based surrogate-assisted particle swarm optimizer [73].…”