Automated Design of Deep Neural Networks - Archive ouverte HAL Access content directly
Journal Articles ACM Computing Surveys Year : 2022

Automated Design of Deep Neural Networks

Abstract

In recent years, research in applying optimization approaches in the automatic design of deep neural networks has become increasingly popular. Although various approaches have been proposed, there is a lack of a comprehensive survey and taxonomy on this hot research topic. In this article, we propose a unified way to describe the various optimization algorithms that focus on common and important search components of optimization algorithms: representation, objective function, constraints, initial solution(s), and variation operators. In addition to large-scale search space, the problem is characterized by its variable mixed design space, it is very expensive, and it has multiple blackbox objective functions. Hence, this unified methodology has been extended to advanced optimization approaches, such as surrogate-based, multi-objective, and parallel optimization.
Not file

Dates and versions

hal-03339896 , version 1 (09-09-2021)

Identifiers

Cite

El-Ghazali Talbi. Automated Design of Deep Neural Networks. ACM Computing Surveys, 2022, 54 (2), pp.1-37. ⟨10.1145/3439730⟩. ⟨hal-03339896⟩
48 View
0 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More