HHaaHMM: A Hyper-Heuristic as a Hidden Markov Model
Willem Van Onsem, Bart Demoen, Patrick De Causmaecker · Lirias · 2014
A new hyper-heuristic algorithm is presented. The approach uses an expectation-maximization strategy in order to find the most likely compressed representation of the hyper-heuristic process. The approach is in principle to calculate a good heuristic at any point in time given the old data.