Lights and Shadows in Evolutionary Deep Learning: Taxonomy, Critical Methodological Analysis, Cases of Study, Learned Lessons, Recommendations and Challenges

Much has been said about the fusion of bio-inspired optimization algorithms\nand Deep Learning models for several purposes: from the discovery of network\ntopologies and hyper-parametric configurations with improved performance for a\ngiven task, to the optimization of the model's parameters as a replacement for\ngradient-based solvers. Indeed, the literature is rich in proposals showcasing\nthe application of assorted nature-inspired approaches for these tasks. In this\nwork we comprehensively review and critically examine contributions made so far\nbased on three axes, each addressing a fundamental question in this research\navenue: a) optimization and taxonomy (Why?), including a historical\nperspective, definitions of optimization problems in Deep Learning, and a\ntaxonomy associated with an in-depth analysis of the literature, b) critical\nmethodological analysis (How?), which together with two case studies, allows us\nto address learned lessons and recommendations for good practices following the\nanalysis of the literature, and c) challenges and new directions of research\n(What can be done, and what for?). In summary, three axes - optimization and\ntaxonomy, critical analysis, and challenges - which outline a complete vision\nof a merger of two technologies drawing up an exciting future for this area of\nfusion research.\n