Learning Instance-wise Sparsity for Accelerating Deep Models

被引:0
|
作者
Liu, Chuanjian [1 ]
Wang, Yunhe [1 ]
Han, Kai [1 ]
Xu, Chunjing [1 ]
Xu, Chang [2 ]
机构
[1] Huawei Noahs Ark Lab, Beijing, Peoples R China
[2] Univ Sydney, FEIT, Sch Comp Sci, Sydney, NSW, Australia
基金
澳大利亚研究理事会;
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Exploring deep convolutional neural networks of high efficiency and low memory usage is very essential for a wide variety of machine learning tasks. Most of existing approaches used to accelerate deep models by manipulating parameters or filters without data, e.g., pruning and decomposition. In contrast, we study this problem from a different perspective by respecting the difference between data. An instance-wise feature pruning is developed by identifying informative features for different instances. Specifically, by investigating a feature decay regularization, we expect intermediate feature maps of each instance in deep neural networks to be sparse while preserving the overall network performance. During online inference, subtle features of input images extracted by intermediate layers of a well-trained neural network can be eliminated to accelerate the subsequent calculations. We further take coefficient of variation as a measure to select the layers that are appropriate for acceleration. Extensive experiments conducted on benchmark datasets and networks demonstrate the effectiveness of the proposed method.
引用
收藏
页码:3001 / 3007
页数:7
相关论文
共 50 条
  • [21] IMAGE-LEVEL SUPERVISED INSTANCE SEGMENTATION USING INSTANCE-WISE BOUNDARY
    Yang, Yuyuan
    Hou, Ya-Li
    Hou, Zhijiang
    Hao, Xiaoli
    Shen, Yan
    2021 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP), 2021, : 1069 - 1073
  • [22] Instance-Wise Minimax-Optimal Algorithms for Logistic Bandits
    Abeille, Marc
    Faury, Louis
    Calauzenes, Clement
    24TH INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE AND STATISTICS (AISTATS), 2021, 130
  • [23] Instance-Wise Dynamic Sensor Selection for Human Activity Recognition
    Yang, Xiaodong
    Chen, Yiqiang
    Yu, Hanchao
    Zhang, Yingwei
    Lu, Wang
    Sun, Ruizhe
    THIRTY-FOURTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THE THIRTY-SECOND INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE AND THE TENTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2020, 34 : 1104 - 1111
  • [24] DRAUC: An Instance-wise Distributionally Robust AUC Optimization Framework
    Dai, Siran
    Xu, Qianqian
    Yang, Zhiyong
    Cao, Xiaochun
    Huang, Qingming
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 36 (NEURIPS 2023), 2023,
  • [25] IQDet: Instance-wise Quality Distribution Sampling for Object Detection
    Ma, Yuchen
    Liu, Songtao
    Li, Zeming
    Sun, Jian
    2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021, 2021, : 1717 - 1725
  • [26] Uncertainty Calibration with Energy Based Instance-Wise Scaling in the Wild Dataset
    Kinn, Mijoo
    Kwon, Junseok
    COMPUTER VISION-ECCV 2024, PT XLVI, 2025, 15104 : 232 - 248
  • [27] Instance-wise or Class-wise? A Tale of Neighbor Shapley for Concept-based Explanation
    Li, Jiahui
    Kuang, Kun
    Li, Lin
    Chen, Long
    Zhang, Songyang
    Shao, Jian
    Xiao, Jun
    PROCEEDINGS OF THE 29TH ACM INTERNATIONAL CONFERENCE ON MULTIMEDIA, MM 2021, 2021, : 3664 - 3672
  • [28] Instance-wise Hard Negative Example Generation for Contrastive Learning in Unpaired Image-to-Image Translation
    Wang, Weilun
    Zhou, Wengang
    Bao, Jianmin
    Chen, Dong
    Li, Houqiang
    2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021), 2021, : 14000 - 14009
  • [29] The advantages of instance-wise reaching definition analyses in Array (S)SA
    Collard, JF
    LANGUAGES AND COMPILERS FOR PARALLEL COMPUTING, 1999, 1656 : 338 - 352
  • [30] AROID: Improving Adversarial Robustness Through Online Instance-Wise Data Augmentation
    Li, Lin
    Qiu, Jianing
    Spratling, Michael
    INTERNATIONAL JOURNAL OF COMPUTER VISION, 2025, 133 (02) : 929 - 950