Evolutionary many-objective optimisation: many once or one many?

By Evan James Hughes. IEEE Congress on Evolutionary Computation, 2005

401 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Theory and Mathematics

View Paper: https://www.semanticscholar.org/paper/e96cba463ef6b71c9323fc655668e9ea86197d8f

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Yaochu Jin
2. Ran Cheng
3. Xing-yi Zhang
4. Ye Tian
5. Kalyanmoy Deb
6. Hisao Ishibuchi
7. Shengxiang Yang
8. Xin Yao

In Computational Theory and Mathematics

1. Yaochu Jin
2. Ran Cheng
3. Xing-yi Zhang
4. Ye Tian
5. Hisao Ishibuchi
6. Xin Yao
7. Kalyanmoy Deb
8. Shengxiang Yang

Citations in the corpus

• Multiobjective Optimization Problems With Complicated Pareto Sets, MOEA/D and NSGA-II (2009, background): “MSOPS uses a number of scalar aggregation functions to guide its search and has demonstrated its advantage for tackling MOPs with many objectives [33], [37], [39].”
• On the Evolutionary Optimization of Many Conflicting Objectives (2007, background): “functions (similar to [11]) and, most interestingly, operators”
• A Decomposition-Based Evolutionary Algorithm for Many Objective Optimization (2015, result): “Similar observations have also been reported in [36].”
• A survey on multi-objective evolutionary algorithms for many-objective problems (2014, methodology): “Hughes [41] used multiple single objective Pareto sampling (MSOPS) [40] for solving many-objective optimization problems.”
• An Evolutionary Many-Objective Optimization Algorithm Using Reference-Point Based Nondominated Sorting Approach, Part II: Handling Constraints and Extending to an Adaptive Approach (2014, methodology): “Although a few many-objective evolutionary algorithms have been suggested in the past [6]–[12], there is still a need for more efficient algorithms for many-objective optimization, similar to popular two or three-objective EMO methods, such as NSGA-II [13], SPEA2 [14], and others.”
• A Survey of Decomposition Methods for Multi-objective Optimization (2014, background): “It is well known that dominance based optimization methods lose their effectiveness with more than 3 objectives [39].”
• Preference-Inspired Coevolutionary Algorithms for Many-Objective Optimization (2013, background): “NSGA-II is known to perform well on bi-objective problems but may experience difficulties in many-objective spaces [5], [6], [8].”
• Controlling Dominance Area of Solutions and Its Impact on the Performance of MOEAs (2007, background): “However, some current research reveals that ranking by Pareto dominance on problems with an increased number of objectives might not longer be effective [3, 4 ,5].”
• Performance of Decomposition-Based Many-Objective Algorithms Strongly Depends on Pareto Front Shapes (2017, background): “The other category is characterized by the use of a scalarizing function for fitness evaluation such as multiple single objective Pareto sampling (MSOPS) [24] and multiobjective evolutionary algorithm based on decomposition (MOEA/D) [25].”
• Shift-Based Density Estimation for Pareto-Based Algorithms in Many-Objective Optimization (2014, background): “, aggregation-based or indicator-based algorithms) are very competitive in manyobjective optimization [33], [38], [71].”
• An Evolutionary Approach to Multiobjective Clustering (2007, background): “PESA-II can also handle any number of objective functions, though its performance (like that of any other Pareto-based MOEA) is likely to deteriorate with larger numbers of objectives [45].”
• Evolutionary many-objective optimization: A short review (2008, background): “Another approach is to use a number of different scalarizing functions for fitness evaluation [9], [10], [20]-[22].”
• Theory of the hypervolume indicator: optimal μ-distributions and the choice of the reference point (2009, background): “In various publications, the reference point is chosen as the nadir point of the investigated solution set, e.g., in [20, 19, 13], while others recommend rescaling of the objective values everytime the hypervolume indicator is computed [25].”