Adopting Harmony Search Algorithm in Deep Learning

A paramount of importance is indicated in the recent decades of research for artificial intelligence and deep learning. The remarkable progress from employing traditional neural network models to adopting artificial intelligence for adaptive learning has significantly enhanced performance in building learning networks. CNN and deep learning have emerged as an indispensable tool for image processing, artificial intelligence analysis, data analytics and many more. The inherent functions of the human brain bypass complexity, streamlining tasks like image segmentation, annotation, and classification, particularly in analysis. The deep learning frameworks are guided by critical hyperparameters, to attain the perfection. A meta-heuristic-based optimization called harmony search algorithm has been introduced in this work to tune up the parameters required to upshot the quality of outputs in variants of CNN.