Fig. 11
From: Learning versus optimal intervention in random Boolean networks
XCS results for the Fission Yeast BN problem. Results are for intervening in the network with a cost per intervention of 2.0. As can be seen, XCS achieves a cost of between 130% and 190% of the optimal cost. Rule set size is significantly higher than optimal, although no post run-compression was performed for the XCS rule sets due to resource constraints