Output details
11 - Computer Science and Informatics
University of Exeter
Visualising High-Dimensional Pareto Relationships in Two-Dimensional Scatterplots
<20> This was presented at EMO, the premier conference for multi-criterion optimisation and builds on a body of our work on visualisation of multi-criterion sets. Novel methods for projecting high dimensional data into two dimensions for visualisation are introduced, which aim to preserve Pareto dominance relationships between data after projection, which, in general, cannot be achieved. We give new guarantees that two of three dominance relationships can be preserved in addition to Pareto shell membership. Thus considerable structure in the original space is preserved. It is also much faster to compute than previous methods making real-time visualisation of optimisers tractable.