Survey on Evolutionary Deep Learning: Principles, Algorithms, Applications and Open Issues

Over recent years, there has been a rapid development of deep learning (DL) in both industry and academia fields. However, finding the optimal hyperparameters DL model often needs high computational cost human expertise. To mitigate above issue, evolutionary computation (EC) as powerful heuristic search approach shown significant merits automated design models, so-called (EDL). This paper aims to analyze EDL from perspective machine (AutoML). Specifically, we firstly illuminate EC regard an optimization problem. According pipeline, systematically introduce methods ranging data preparation, generation, deployment with new taxonomy (i.e., what how evolve/optimize), focus on discussions solution representation paradigm handling problem by EC. Finally, key applications, open issues potentially promising lines future research are suggested. survey reviewed developments offers insightful guidelines for EDL.