Bacterial foraging optimization with double role of reproduction and step adaptation
Mohamed Skander Daas, Salim Chıkhı, Mohamed Batouche · 2015
Bacterial foraging optimization (BFO) is a swarm intelligence algorithm inspired by the foraging behavior of Escherichia coli (E.coli). BFO has demonstrated a great performance for many real world problems. But it has two noteworthy disadvantages: its large number of parameters and its fixed step-size unit. In this work, we present a new effective version of BFO algorithm. At first the Elimination-Dispersal step and some parameters are discarded. The benefits of the Elimination-Dispersal step are rewarded on the new proposed reproduction and step-adaptation mechanisms. These two mechanisms are very simple and effective with little computational cost. Unlike original BFO reproduction, the proposed reproduction mechanism allows new born bacteria to take different positions from their clones. The resulting bacteria distribution keeps a good Elitism / Diversity trade-off. In the same context, the proposed step-adaptation mechanism use small and big step-sizes which serves exploitation as well as exploration. Performances of the proposed algorithm are compared to many other results obtained from the literature. Results show that the proposed algorithm outperforms significantly almost compared algorithms on many test functions.