Grammatical Evolution-Driven Algorithm for Efficient and Automatic Hyperparameter Optimisation of Neural Networks

Neural networks have revolutionised the way we approach problem solving across multiple domains; however, their effective design and efficient use of computational resources is still a challenging task. One most important factors influencing this process model hyperparameters which vary significantly with models datasets. Recently, there has been an increased focus on automatically tuning these to reduce complexity optimise resource utilisation. From traditional human-intuitive methods random search, grid Bayesian optimisation, evolutionary algorithms, significant advancements made in direction that promise improved performance while using fewer resources. In article, propose HyperGE, two-stage for driven by grammatical evolution (GE), bioinspired population-based machine learning algorithm. GE provides advantage it allows users define own grammar generating solutions, making ideal defining search spaces datasets models. We test HyperGE fine-tune VGG-19 ResNet-50 pre-trained three benchmark demonstrate space reduced factor ~90% Stage 2 number trials. could become invaluable tool within deep community, allowing practitioners greater freedom when exploring complex domains hyperparameter fine-tuning.