Fig. 12
From: Learning versus optimal intervention in random Boolean networks
XCS relative performance with a cost of 2.0 per intervention link. Each point represents a single network instance given parameter settings of N and K. The horizontal position of each point represents the relative performance of XCS versus the optimal rule calculator operating on the network instance in question (i.e. % of optimal cost). The vertical position of each point represents the relative performance of XCS versus the optimal rule calculator operating on the network instance in question. Note that the y-axis is % of rule set size created via the optimal rule calculator but is not necessarily % of optimal rule set size