Introduction to Neural Architecture Search (Reinforcement Learning approach)

Author: Hamdi M Abed

AI that creates AI, that is what first people described automated machine learning (AutoML). But is it really like that? AutoML gained notable attention since 2016 when Google brain revealed their first Neural Architecture Search (NAS) publication by Quac Le and Barret Zoph , which was used in several applications. The idea is to generate some candidate children networks, train them, optimize it’s accuracy until successfully select the best architecture for some certain task. In…