Multi-Objective Quality Evaluation

A key issue in multi-objective optimisation is how to evaluate and compare solution sets obtained by various optimisation algorithms. This is not a trivial task. The outcome of multi-objective optimisers is typically a set of incomparable solutions (i.e. being Pareto nondominated to each other), and using a scalar value to reflect the quality of the whole solution set requires careful design/selection of evaluation indicators.

• Survey and Critical Review

• [CSUR19] A survey of 100 quality indicators in multi-objective optimisation, providing guidance on how to select and use quality indicators in various situations. [Read More]

• [TSE20, ICSE18] A critical review and methodological guidance as to how to select and use quality evaluation methods in different multi-objective software engineering scenarios. [Read More]

• Quality Indicator Design

  • • [TEVC21] A kernel-based quality indicator which is able to reflect convergence and diversity as well as distribution of solution sets.

    • [GECCO15] A quality indicator to compare relative quality (convergence and diversity) of solution sets with various Pareto fronts and any objective dimensionality.

    • [TCYB14] A quality indicator to compare the diversity of solution sets with various Pareto fronts and any objective dimensionality.