On measuring multiobjective evolutionary algorithm performance

By D. V. Veldhuizen, Gary B. Lamont. Proceedings of the 2000 Congress on Evolutionary Computation. CEC00 (Cat. No.00TH8512), 2000

550 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Theory and Mathematics

View Paper: https://www.semanticscholar.org/paper/9854d7eb5ae0a363074b13d5e88a6d437dc67826

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

In Computational Theory and Mathematics

Citations in the corpus

Consistencies and Contradictions of Performance Metrics in Multiobjective Optimization (2014, background): “For instance, [41] defined three metrics:”
An improved NSGA-II algorithm for multi-objective lot-streaming flow shop scheduling problem (2014, methodology): “The |S*| is the number of reference solutions in S*. 4.4 The spread of non-dominated solutions The SC(Sj) is utilised to measure the performance of non-dominated solutions in distribution in the objective space (Veldhuizen and Lamont 2000).”
Preference-Inspired Coevolutionary Algorithms for Many-Objective Optimization (2013, methodology): “To further understand the performance of the algorithms, we have also separately calculated proximity (as measured by generational distance (GD) [51]) and diversity (as measured by the spread metric [2]) measures for the seven-objective WFG4–9 benchmark functions.”
Performance assessment of multiobjective optimizers: an analysis and review (2003, background): “Van Veldhuizen and Lamont [2], for instance, applied a combination of three indicators, where”
Multiobjective Immune Algorithm with Nondominated Neighbor-Based Selection (2008, unknown): “Similar metrics were proposed by Schott (1995), Rudolph (1998), Zitzler et al. (2000) and Van Veldhuizen and Lamont (2000) et al. Spacing : Let A be the ﬁnal approximate Pareto-optimal set obtained by an MOEA. where ¯ d is the average value of all d i , and k is the number of objective functions.”
Controlling Dominance Area of Solutions and Its Impact on the Performance of MOEAs (2007, background): “IGD takes the average distance for all members in the true Pareto front to their nearest solutions in the obtained set of non-dominated solutions (exactly the inverse process followed by Generational Distance GD [ 14 ]).”
A Competitive-Cooperative Coevolutionary Paradigm for Dynamic Multiobjective Optimization (2009, background): “Many comparative studies [11], [29], [57], [63] made use of a suite of unary performance metrics [68] pertinent to the MO optimization goals of proximity, diversity, and distribution.”
A Tutorial on Evolutionary Multiobjective Optimization (2004, methodology): “, the following combination of unary quality measures used in [35]: average distance to the Pareto front, diversity of the Pareto front approximation, and the number of vectors in the Pareto front approximation.”
Techniques for highly multiobjective optimisation: some nondominated points are better than others (2007, result): “ests as statistically significant if one or other algorithm was the winner suitably often. This is not ideal or authoritative (there is a growing literature on performance metrics for MOEAs, e.g. see [10, 26, 28]); however, early investigations showed that, in the case of the experiments done here, ”