Output details
11 - Computer Science and Informatics
University of Exeter
Visualizing Mutually Nondominating Solution Sets in Many-Objective Optimization
<12> This paper appears in one of the top ranked Computer Science and Artificial Intelligence journals (second in the Computer Science, Artificial Intelligence & Computer Science and Theory & Methods categories). The paper continues a line of investigation on how to display and interpret high-dimensional mutually non-dominating sets, which result from many-objective optimisation problems. Spectral seriation is used for the clear presentation of heatmaps and it introduces novel dimension reduction methods, while preserving dominance relations. The most significant of these relies on a new metric, the dominance distance, which for the first time, allows principled dimension reduction.