Abstract
This paper proposes a novel implementation of micro-Differential Evolution (μDE) that incorporates within the DE scheme an extra search move that attempts to improve the best solution by perturbing it along the axes. These extra moves complement the DE search logic and allows the exploration of the decision space from an alternative perspective. In addition, these extra moves at subsequent activations tend to explore a progressively narrowing area. This mechanism increases the exploitation of the original scheme thus helping μDE to prevent from stagnation. An experimental set-up including various test problems and dimensionality values has been considered. Numerical results show that the proposed algorithm enhances upon the original μDE performance and, despite its simplicity, is competitive with modern complex DE based algorithms.