An Evolutionary Many-Objective Optimization Algorithm Using Reference-Point-Based Nondominated Sorting Approach, Part I: Solving Problems With Box Constraints
Citations
1,247 citations
Cites background or methods from "An Evolutionary Many-Objective Opti..."
...We also propose the use of PBI metric (instead of the Tchebycheff metric), as PBI metric was found to work better in the original study [1]....
[...]
...The number of reference points, population size, and other parameters are kept in agreement with the original study [1] and are tabulated in Tables I and II....
[...]
...Another aspect of the extension is that if all population members are feasible or an unconstrained problem is supplied, the constrained NSGA-III reduces to the original unconstrained NSGA-III algorithm [1]....
[...]
...The rest of the NSGA-III procedure described in the original paper [1] remains the same....
[...]
...Notice how the procedure becomes similar to the unconstrained NSGA-III selection operator (described in the original study [1]) when there is no infeasible population member or when there are no equality or inequality constraints specified in the optimization problem formulation....
[...]
1,072 citations
1,020 citations
Cites background or methods from "An Evolutionary Many-Objective Opti..."
...For example, in NSGA-III [39], a set of reference pointsor r ference linesare used for niche preservation to manage diversity in each subspace for many-objective optimizatio n, which effectively enhances convergence by giving priority to solutions closer to the reference points....
[...]
...As already shown in [39], reference points can also be used to generate a subset of preferred Pareto optimal solutions, although NSGA-III can be seen as a decomposition based approach if the reference points are evenly distributed in t he whole objective space....
[...]
...Since weight vectorsare typically used to denote the importance of objectives in weighted aggregation, differe nt t rminologies have been coined in the second type of decomposition based approaches to refer to vectors that decompos e he original objective space, including direction vectors[29], reference lines[39], and reference vectors[40]....
[...]
..., the one adopted in NSGAIII [39], the proposed angle normalization approach has two major differences: (1) normalizing the angles (instead of t he objectives) will not change the actual positions of the cand idate solutions, which is important convergence information for the proposed RVEA; (2) angle normalization, which is independently carried out inside each subspace, does not influence t he distribution of the candidate solutions in other subspaces ....
[...]
...The scaling approach is recommended in [39], where each objective is multiplied by a coefficient pi−1, where p is a parameter that controls the scaling size andi = 1, ....
[...]
915 citations
900 citations
Cites background or methods from "An Evolutionary Many-Objective Opti..."
...Furthermore, the recently proposed NSGA-III [49] also employs a decomposition-based idea to...
[...]
...binary crossover (SBX) [61] and polynomial mutation [62] as in [49]....
[...]
...In this paper, we use the penalty-based boundary intersection (PBI) approach [41], due to its promising performance for many-objective optimization reported in [49]....
[...]
...As discussed in [49], in order to have intermediate weight vectors within the simplex, we should set H ≥ m....
[...]
...Most existing studies of MOEA/D in many-objective scenario mainly concentrate on investigations of its search behavior (see [45]–[49])....
[...]
References
37,111 citations
"An Evolutionary Many-Objective Opti..." refers background or methods in this paper
...Thereafter, we propose a new method that uses the framework of NSGA-II procedure [5], but works with a set of supplied or predefined reference points and demonstrates its efficacy in solving two-objective to 15-objective optimization problems....
[...]
...The basic framework of the proposed many-objective NSGA-II (or NSGA-III) is similar to the original NSGA-II algorithm [5] with significant changes in its selection operator....
[...]
...Second, implementation of a diversity-preservation operator (such as the crowding distance operator [5] or clustering operator [6]) becomes a computationally expensive operation....
[...]
12,134 citations
7,512 citations
"An Evolutionary Many-Objective Opti..." refers background in this paper
...For this purpose, the performance metrics (such as hyper-volume measure [7] or other metrics [3], [8]) are either computationally too expensive or may not be meaningful....
[...]
6,657 citations
"An Evolutionary Many-Objective Opti..." refers background or methods or result in this paper
...Such a distribution was also reported in the original MOEA/D study [10]....
[...]
...MOEA/D [10] uses a predefined set of weight vectors to maintain a diverse set of trade-off solutions....
[...]
...A recently proposed MOEA/D procedure [10] uses this concept....
[...]
...This observation is similar to that concluded in the original MOEA/D study [10] based on two-objective and three-objective problems....
[...]
...We compare the performance of the proposed NSGA-III with two versions of an existing many-objective EMO (MOEA/D [10]), as the method is somewhat similar to the proposed...
[...]
4,976 citations
"An Evolutionary Many-Objective Opti..." refers background in this paper
...The use of fixed conedomination [33], [34] or variable cone-domination [35] principles can also be tried....
[...]