BlockQNN: Efficient Block-Wise Neural Network Architecture Generation

doi:10.1109/TPAMI.2020.2969193

CASIA OpenIR > 多模态人工智能系统全国重点实验室 > 模式分析与学习

	BlockQNN: Efficient Block-Wise Neural Network Architecture Generation
	Zhong, Zhao1 ; Yang, Zichen 2; Deng, Boyang 2; Yan, Junjie 2; Wu, Wei 2; Shao, Jing 2; Liu, Cheng-Lin3,4
发表期刊	IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
ISSN	0162-8828
	2021-07-01
卷号	43 期号:7 页码:2314-2328
通讯作者	Liu, Cheng-Lin(liucl@nlpr.ia.ac.cn)
摘要	Convolutional neural networks have gained a remarkable success in computer vision. However, most popular network architectures are hand-crafted and usually require expertise and elaborate design. In this paper, we provide a block-wise network generation pipeline called BlockQNN which automatically builds high-performance networks using the Q-Learning paradigm with epsilon-greedy exploration strategy. The optimal network block is constructed by the learning agent which is trained to choose component layers sequentially. We stack the block to construct the whole auto-generated network. To accelerate the generation process, we also propose a distributed asynchronous framework and an early stop strategy. The block-wise generation brings unique advantages: (1) it yields state-of-the-art results in comparison to the hand-crafted networks on image classification, particularly, the best network generated by BlockQNN achieves 2.35 percent top-1 error rate on CIFAR-10. (2) it offers tremendous reduction of the search space in designing networks, spending only 3 days with 32 GPUs. A faster version can yield a comparable result with only 1 GPU in 20 hours. (3) it has strong generalizability in that the network built on CIFAR also performs well on the larger-scale dataset. The best network achieves very competitive accuracy of 82.0 percent top-1 and 96.0 percent top-5 on ImageNet.
关键词	Computer architecture Task analysis Neural networks Network architecture Graphics processing units Acceleration Indexes Convolutional neural network neural architecture search AutoML reinforcement learning Q-learning
DOI	10.1109/TPAMI.2020.2969193
收录类别	SCI
语种	英语
资助项目	Major Project for New Generation of AI[2018AAA0100400] ; National Natural Science Foundation of China (NSFC)[61721004] ; National Natural Science Foundation of China (NSFC)[61633021]
项目资助者	Major Project for New Generation of AI ; National Natural Science Foundation of China (NSFC)
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
WOS记录号	WOS:000692540900011
出版者	IEEE COMPUTER SOC
七大方向——子方向分类	模式识别基础
引用统计	被引频次：61[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/45739
专题	多模态人工智能系统全国重点实验室_模式分析与学习
通讯作者	Liu, Cheng-Lin
作者单位	1.Univ Chinese Acad Sci, Inst Automat, Chinese Acad Sci, NLPR, Beijing 100190, Peoples R China 2.Sensetime Res Inst, SenseTime Grp Ltd, Beijing, Peoples R China 3.Chinese Acad Sci, Inst Automat, NLPR, Beijing, Peoples R China 4.Univ Chinese Acad Sci, CAS Ctr Excellence Brain Sci & Intelligence, Beijing 100190, Peoples R China
第一作者单位	模式识别国家重点实验室
通讯作者单位	模式识别国家重点实验室
推荐引用方式 GB/T 7714	Zhong, Zhao,Yang, Zichen,Deng, Boyang,et al. BlockQNN: Efficient Block-Wise Neural Network Architecture Generation[J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,2021,43(7):2314-2328.
APA	Zhong, Zhao.,Yang, Zichen.,Deng, Boyang.,Yan, Junjie.,Wu, Wei.,...&Liu, Cheng-Lin.(2021).BlockQNN: Efficient Block-Wise Neural Network Architecture Generation.IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,43(7),2314-2328.
MLA	Zhong, Zhao,et al."BlockQNN: Efficient Block-Wise Neural Network Architecture Generation".IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 43.7(2021):2314-2328.