Abstract
Compact algorithms are Estimation of Distribution Algorithms which mimic the behavior of population-based algorithms by means of a probabilistic representation of the population of candidate solutions. Compared to an actual population, a probabilistic model requires a much smaller memory, which allows algorithms with limited memory footprint. This feature is extremely important in some engineering applications, e.g. robotics and real-time control systems. This paper proposes a compact implementation of Bacterial Foraging Optimization (cBFO). cBFO employs the same chemotaxis scheme of population-based BFO, but without storing a swarm of bacteria. Numerical results, carried out on a broad set of test problems with different dimensionalities, show that cBFO, despite its minimal hardware requirements, is competitive with other memory saving algorithms and clearly outperforms its population-based counterpart.