ARCHIVES

This paper introduces the Musca Optimizer, a novel gradient descent optimization algorithm inspired by the persistent behavior of Musca domestica (common house fly). When the optimizer detects a potential local minimum (near-zero gradient), it introduces a random perturbation to escape from the current position. If the algorithm repeatedly returns to the same point after multiple perturbations, it concludes that this point represents a robust solution possibly a global minimum or the best reachable local minimum within the search space. This work provides mathematical formulation, pseudocode, theoretical analysis, and experimental validation on multi-modal test functions.