Citation Context

...lti-objective optimizationsproblems by using evolutionary algorithms [12],smulti-objective evolutionary algorithms (MOEAs) havesbeen much researched and are widely used to solvesnumerous applications =-=[13]-=-[14] in recent years. MOEAssbenefit from the evolutionary algorithm’s ability to generatesa set of solutions concurrently in a single run therebysyielding several trade-off solutions.sDifferential Evo...

Citation Context

...objective optimizationsproblems by using evolutionary algorithms [12],smulti-objective evolutionary algorithms (MOEAs) havesbeen much researched and are widely used to solvesnumerous applications [13]=-=[14]-=- in recent years. MOEAssbenefit from the evolutionary algorithm’s ability to generatesa set of solutions concurrently in a single run therebysyielding several trade-off solutions.sDifferential Evoluti...

Citation Context

...ONsANY real worldsproblems can be formulated assoptimization problems with multiplesobjectives. Sincesthe first attempt to solve multi-objective optimizationsproblems by using evolutionary algorithms =-=[12]-=-,smulti-objective evolutionary algorithms (MOEAs) havesbeen much researched and are widely used to solvesnumerous applications [13][14] in recent years. MOEAssbenefit from the evolutionary algorithm’s...

Citation Context

...nt years. MOEAssbenefit from the evolutionary algorithm’s ability to generatesa set of solutions concurrently in a single run therebysyielding several trade-off solutions.sDifferential Evolution (DE) =-=[8]-=- is one of the mostscommonly used EAs. It is a simple and powerfulspopulation-based stochastic direct search method for solvingsnumerical optimization problems in continuous search space.sIn DE, one c...

Citation Context

... be learned and applied to efficiently evolvesthe population at different stages. The SaDE algorithm hassdemonstrated promising performance when solving differentstypes of optimization problems [1][2]=-=[3]-=-[5].sIn this work, we improve the MOSaDE algorithm [5] withsV. L. Huang, S. Zhao, R. Mallipeddi and P. N. Suganthan are with School ofsElectrical and Electronic Engineering, Nanyang Technological Univ...

Citation Context

...rs can be learned and applied to efficiently evolvesthe population at different stages. The SaDE algorithm hassdemonstrated promising performance when solving differentstypes of optimization problems =-=[1]-=-[2][3][5].sIn this work, we improve the MOSaDE algorithm [5] withsV. L. Huang, S. Zhao, R. Mallipeddi and P. N. Suganthan are with School ofsElectrical and Electronic Engineering, Nanyang Technologica...

Citation Context

...xclusivesintegers randomly generated within the range [1, NP], thesesindices are randomly generated once for each mutant vector.sThese mutation strategies have been commonly used whenssolving MOPs [4]=-=[6]-=-[7][10][11].sThe main steps of OW-MOSaDE are described below:sStep 1. Randomly initialize a population of NP individuals.sInitialize n_obj number of vectors each corresponding to onesobjective. Each v...

Citation Context

...ngsfactor F was generated for each variable from a Gaussiansdistribution (0,s1)N [7]. Zaharie proposed a parametersadaptation for DE (ADE) based on the concept of controllingsthe population diversity =-=[10]-=-. Following the same ideas,sZaharie and Petcu designed an adaptive Pareto DE algorithmsfor solving MOPs and analyzed its parallel implementations[11]. Xue et al. used a fuzzy logic controller (FLC) to...

Citation Context

...he concept of controllingsthe population diversity [10]. Following the same ideas,sZaharie and Petcu designed an adaptive Pareto DE algorithmsfor solving MOPs and analyzed its parallel implementations=-=[11]-=-. Xue et al. used a fuzzy logic controller (FLC) tosdynamically adjust the parameters of multi-objectivesdifferential evolution [9].sIII. OBJECTIVE-WISE LEARNING IN MULTI-OBJECTIVEsSADEsThe earlier Mu...

Citation Context

...y exclusivesintegers randomly generated within the range [1, NP], thesesindices are randomly generated once for each mutant vector.sThese mutation strategies have been commonly used whenssolving MOPs =-=[4]-=-[6][7][10][11].sThe main steps of OW-MOSaDE are described below:sStep 1. Randomly initialize a population of NP individuals.sInitialize n_obj number of vectors each corresponding to onesobjective. Eac...

Citation Context

...f PDE, bysencoding the crossover rate into each individual andssimultaneously evolving with other parameters. The scalingsfactor F was generated for each variable from a Gaussiansdistribution (0,s1)N =-=[7]-=-. Zaharie proposed a parametersadaptation for DE (ADE) based on the concept of controllingsthe population diversity [10]. Following the same ideas,sZaharie and Petcu designed an adaptive Pareto DE alg...

Citation Context

...algorithmsfor solving MOPs and analyzed its parallel implementations[11]. Xue et al. used a fuzzy logic controller (FLC) tosdynamically adjust the parameters of multi-objectivesdifferential evolution =-=[9]-=-.sIII. OBJECTIVE-WISE LEARNING IN MULTI-OBJECTIVEsSADEsThe earlier Multi-objective SaDE algorithm (MOSaDE)s[5] is an extension of our recently developed SaDE [3] tosoptimize problems with multiple obj...

Citation Context

...l.ntu.edu.sg,sepnsugan@ntu.edu.sg)sobjective-wise learning strategies (called as OW-MOSaDE)sto solve problems with multiple conflicting objectives andsevaluate the performance on the 13 test problems =-=[15]-=-. Thesoriginal MOSaDE learns just one set of parameters for all thesobjectives. In MOPs, different objective functions mayspossess different properties. Hence, it can be beneficial toslearn one set of...