Neural Architecture Search: A Survey
This is a survey paper that summarizes incremental research in neural architecture search for machine learning practitioners.
The paper surveys automated neural architecture search methods, addressing the time-consuming and error-prone manual design of neural architectures by categorizing existing work based on search space, search strategy, and performance estimation strategy.
Deep Learning has enabled remarkable progress over the last years on a variety of tasks, such as image recognition, speech recognition, and machine translation. One crucial aspect for this progress are novel neural architectures. Currently employed architectures have mostly been developed manually by human experts, which is a time-consuming and error-prone process. Because of this, there is growing interest in automated neural architecture search methods. We provide an overview of existing work in this field of research and categorize them according to three dimensions: search space, search strategy, and performance estimation strategy.