Implementing the cumulative difference plot in the IOHanalyzer

Etor Arza, Josu Ceberio, Ekhiñe Irurozki, Aritz Pérez

Abstract: The IOHanalyzer is a web-based framework that enables an easy visualization and comparison of the quality of stochastic optimization algorithms. IOHanalyzer offers several graphical and statistical tools analyze the results of such algorithms. In this work, we implement the cumulative difference plot in the IOHanalyzer. The cumulative difference plot [1] is a graphical approach that compares two samples through the first-order stochastic dominance. It improves upon other graphical approaches with the ability to distinguish between a small magnitude of difference and high uncertainty.