A multi-directional modified Physarum algorithm for optimal multi-objective discrete decision making

Masi, Luca and Vasile, Massimiliano (2013) A multi-directional modified Physarum algorithm for optimal multi-objective discrete decision making. In: EVOLVE. Studies in Computational Intelligence Series . Springer Berlin Heidelberg, Berlin. ISBN 978-3-642-32725-4

[img] PDF
Masi_L_Vasile_M_Pure_A_multi_directional_modified_Physarum_algorithm_for_optimal_multi_objective_discrete_decision_making_Jun_2013.pdf
Preprint

Download (695kB)

    Abstract

    This paper will address an innovative bio-inspired algorithm able to incrementally grow decision graphs in multiple directions for discrete multi-objective optimisation. The algorithm takes inspiration from the slime mould Physarum Polycephalum, an amoeboid organism that in its plasmodium state extends and optimizes a net of veins looking for food. The algorithm is here used to solve multi-objective Traveling Salesman and Vehicle Routing Problems selected as representative examples of multi-objective discrete decision making problems. Simulations on selected test cases showed that building decision sequences in two directions and adding a matching ability (multi-directional approach) is an advantageous choice if compared with the choice of building decision sequences in only one direction (unidirectional approach). The ability to evaluate decisions from multiple directions enhances the performance of the solver in the construction and selection of optimal decision sequences.