A Survey on Neural Architecture Search Based on Reinforcement Learning

TL;DR

This survey reviews reinforcement learning-based neural architecture search methods, highlighting their development, improvements, and variants aimed at optimizing neural network structures automatically for various tasks.

Contribution

It provides a comprehensive overview of reinforcement learning approaches in neural architecture search, including recent advancements and adaptations for complex and resource-constrained environments.

Findings

Reinforcement learning effectively automates neural architecture design.

Recent methods improve search efficiency and adaptability.

Variants address complex structures and limited resources.

Abstract

The automation of feature extraction of machine learning has been successfully realized by the explosive development of deep learning. However, the structures and hyperparameters of deep neural network architectures also make huge difference on the performance in different tasks. The process of exploring optimal structures and hyperparameters often involves a lot of tedious human intervene. As a result, a legitimate question is to ask for the automation of searching for optimal network structures and hyperparameters. The work of automation of exploring optimal hyperparameters is done by Hyperparameter Optimization. Neural Architecture Search is aimed to automatically find the best network structure given specific tasks. In this paper, we firstly introduced the overall development of Neural Architecture Search and then focus mainly on providing an overall and understandable survey about Neural Architecture Search works that are relevant with reinforcement learning, including improvements and variants based on the hope of satisfying more complex structures and resource-insufficient environment.