Thomson SL & Ochoa G (2020) The Local Optima Level in Chemotherapy Schedule Optimisation. In: Zarges C & Paquete L (eds.) Evolutionary Computation in Combinatorial Optimization. EvoCOP 2020. Lecture Notes in Computer Science, 12102. EvoCOP 2020: European Conference on Evolutionary Computation in Combinatorial Optimization, Seville, Spain, 15.04.2020-17.04.2020. Cham, Switzerland: Springer, pp. 197-213. https://doi.org/10.1007/978-3-030-43680-3_13
Abstract In this paper a multi-drug Chemotherapy Schedule Optimisation Problem (CSOP) is subject to Local Optima Network (LON) analysis. LONs capture global patterns in fitness landscapes. CSOPs have not previously been subject to fitness landscape analysis. We fill this gap: LONs are constructed and studied for meaningful structure. The CSOP formulation presents novel challenges and questions for the LON model because there are infeasible regions in the fitness landscape and an unknown global optimum; it also brings a topic from healthcare to LON analysis. Two LON Construction algorithms are proposed for sampling CSOP fitness landscapes: a Markov-Chain Construction Algorithm and a Hybrid Construction Algorithm. The results provide new insight into LONs of highly-constrained spaces, and into the proficiency of search operators on the CSOP. Iterated Local Search and Memetic Search, which are the foundations for the LON algorithms, are found to markedly out-perform a Genetic Algorithm from the literature.
Keywords Combinatorial fitness landscapes; Local Optima Networks; Search space analysis