A review of neural architecture search

被引：71
作者：
Baymurzina, Dilyara [1 ]
Golikov, Eugene [1 ,3 ]
Burtsev, Mikhail [1 ,2 ]
机构：
[1] Moscow Inst Phys & Technol, Neural Networks & Deep Learning Lab, 9 Inst Skiy Lane, Dolgoprudnyi 141701, Moscow Region, Russia
[2] Artificial Intelligence Res Inst AIRI, Moscow, Russia
[3] Ecole Polytech Fed Lausanne, Route Cantonale, CH-1015 Lausanne, Vaud, Switzerland
来源：
NEUROCOMPUTING | 2022年 / 474卷
关键词：
Neural architecture search; NAS; AutoML; EVOLUTION; NETWORKS;
D O I：
10.1016/j.neucom.2021.12.014
中图分类号：
TP18 [人工智能理论];
学科分类号：
081104 ; 0812 ; 0835 ; 1405 ;
摘要：
Despite the impressive progress in neural network architecture design, improving the performance of the existing state-of-the-art models has become increasingly challenging. For this reason, the paradigm for neural architecture design is shifting from being expert-driven to almost fully automated. An emerging body of research related to such machine-aided design is called a Neural Architecture Search (NAS). This paper reviews the recent works on NAS and highlights several crucial concepts and problems of this field. (c) 2021 Elsevier B.V. All rights reserved.
引用
收藏
页码：82 / 93
页数：12
相关论文
共 86 条
[1]  
[Anonymous], 2020, ARXIV PREPRINT ARXIV
[2]  
[Anonymous], 2016, Hypernetworks
[3]  
[Anonymous], 2019, EfficientNet-EdgeTPU: Creating Accelerator-Optimized Neural Networks with AutoML
[4]  
Baker B., 2016, P 5 INT C LEARNING R
[5]  
Bender G, 2018, PR MACH LEARN RES, V80
[6]  
Cai H, 2018, INT C LEARN REPR
[7]  
Cai H, 2018, AAAI CONF ARTIF INTE, P2787
[8]   Progressive Differentiable Architecture Search: Bridging the Depth Gap between Search and Evaluation [J].
Chen, Xin ;
Xie, Lingxi ;
Wu, Jun ;
Tian, Qi .
2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2019), 2019, :1294-1303
[9]  
Chu X., 2020, ARXIV200503566
[10]   Scaling MAP-Elites to Deep Neuroevolution [J].
Colas, Cedric ;
Huizinga, Joost ;
Madhavan, Vashisht ;
Clune, Jeff .
GECCO'20: PROCEEDINGS OF THE 2020 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE, 2020, :67-75
123456789