Evolving Deep Neural Network by Customized Moth-Flame Optimization Algorithm for Underwater Targets Recognition

Enhancing underwater target localization through proximity-driven recurrent neural networks

“…Khishe et al [ 27 ], proposed the use of the Moth Flame Optimization (MFO) algorithm to fine-tune Deep Neural Networks for recognizing various underwater sonar datasets, while acknowledging the common challenges that metaheuristic algorithms face, such as premature convergence, local minima entrapment, and failure to converge within reasonable timeframes, particularly in high-dimensional search spaces. It emphasizes the critical significance of spiral flight within the MFO, which determines how moths modify their positions relative to flames, managing the transition between the exploration and exploitation stages.…”