PHD-NAS: Preserving helpful data to promote Neural Architecture Search

doi:10.1016/j.neucom.2024.127646

	PHD-NAS: Preserving helpful data to promote Neural Architecture Search
	Lu, Shun 1,2; Hu, Yu 1,2; Yang, Longxing 1,2; Mei, Jilin 1; Sun, Zihao 1,2; Tan, Jianchao 3; Song, Chengru 3
	2024-06-28
发表期刊	NEUROCOMPUTING
ISSN	0925-2312
卷号	587 页码:14
摘要	Neural Architecture Search (NAS) has achieved promising results in many domains. However, the enormous computational burden consumed by the NAS procedure significantly hinders its application. Existing works focus on mitigating the search cost by either designing a more efficient algorithm or searching in an elaborately designed search space, heavily relying on expert experience and domain knowledge. We notice that few works focus on dataset optimization for NAS, however, the truth is that not all samples are essential for the search process, which can be omitted actually. Therefore, we propose to only preserve helpful data for the supernet training to improve the efficiency. Specifically, we compute the forgetting and remembering events for each sample during the supernet training to determine the data importance. Samples that the supernet has predicted correctly in consecutive epochs have low importance and will be gradually removed from the dataset during training. We further formulate our method into a unified cycled -learning framework for jointly optimizing proxy dataset and architecture search. By combining with different algorithms, we demonstrate that our framework can find architectures with comparable performance using much less training data and search time in various search spaces and benchmarks, validating the effectiveness of our method.
关键词	Neural architecture search Dataset optimization Forgetting events and remembering events
DOI	10.1016/j.neucom.2024.127646
收录类别	SCI
语种	英语
资助项目	National Key R&D Program of China[2018AAA0102701] ; National Natural Science Foundation of China[62176250]
WOS研究方向	Computer Science
WOS类目	Computer Science, Artificial Intelligence
WOS记录号	WOS:001231918800001
出版者	ELSEVIER
引用统计	被引频次：1[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/40085
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Hu, Yu
作者单位	1.Chinese Acad Sci, Inst Comp Technol, Res Ctr Intelligent Comp Syst, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Sch Comp Sci & Technol, Beijing 100049, Peoples R China 3.Kuaishou Technol, Beijing 100085, Peoples R China
推荐引用方式 GB/T 7714	Lu, Shun,Hu, Yu,Yang, Longxing,et al. PHD-NAS: Preserving helpful data to promote Neural Architecture Search[J]. NEUROCOMPUTING,2024,587:14.
APA	Lu, Shun.,Hu, Yu.,Yang, Longxing.,Mei, Jilin.,Sun, Zihao.,...&Song, Chengru.(2024).PHD-NAS: Preserving helpful data to promote Neural Architecture Search.NEUROCOMPUTING,587,14.
MLA	Lu, Shun,et al."PHD-NAS: Preserving helpful data to promote Neural Architecture Search".NEUROCOMPUTING 587(2024):14.