Informed mutation operator using machine learning for optimization in epidemics prevention
AbstractThis paper concerns using evolutionary algorithms for optimization in epidemics prevention. Spreading of the epidemic is simulated on a graph and the goal of optimization is to determine which vertices in the graph to vaccinate in order to minimize the number of vertices affected by the epidemic. Decisions whether to vaccinate a vertex or not are represented using a binary genotype. In this paper an informed mutation operator is introduced. The presented operator uses machine learning approach for determining which positions in the genotype to flip. The learning model used in the paper is a neural network trained using graph-based features of the vertices (such as a vertex degree) and the information how often on average a given vertex is infected. Once trained, the learning model helps determine which positions in the current solution to mutate. Results presented in the paper suggest that the proposed informed operator improves the ability of the evolutionary algorithm to produce good solutions to the tackled problem. Interestingly, a very good generalization was achieved. The model built using problem instances with 1000 vertices in the graph improved solutions for problem instances with up to 20000 vertices
|Publication size in sheets||0.5|
|Book||Takadama Keiki, Aguirre Hernan (eds.): Proceedings of the Genetic and Evolutionary Computation Conference (GECCO'18), 2018, Association for Computing Machinery, ISBN 9781450356183, 1546 p.|
|Publication indicators||= 2|
* presented citation count is obtained through Internet information analysis and it is close to the number calculated by the Publish or Perish system.