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
  • [41] Instance-Wise Weighted Nonnegative Matrix Factorization for Aggregating Partitions with Locally Reliable Clusters
    Zheng, Xiaodong
    Zhu, Shanfeng
    Gao, Junning
    Mamitsuka, Hiroshi
    PROCEEDINGS OF THE TWENTY-FOURTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE (IJCAI), 2015, : 4091 - 4097
  • [42] Hardness vs Randomness, Revised: Uniform, Non-Black-Box, and Instance-Wise
    Chen, Lijie
    Tell, Roei
    2021 IEEE 62ND ANNUAL SYMPOSIUM ON FOUNDATIONS OF COMPUTER SCIENCE (FOCS 2021), 2022, : 125 - 136
  • [43] Instance-Wise Domain Generalization for Cross-Scene Wetland Classification With Hyperspectral and LiDAR Data
    Guo, Fangming
    Li, Zhongwei
    Ren, Guangbo
    Wang, Leiquan
    Zhang, Jie
    Wang, Jianbu
    Hu, Yabin
    Yang, Min
    IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2025, 63
  • [44] Accelerating Sparse Convolution with Column Vector-Wise Sparsity
    Tan, Yijun
    Han, Kai
    Zhao, Kang
    Yu, Xianzhi
    Du, Zidong
    Chen, Yunji
    Wang, Yunhe
    Yao, Jun
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 35, NEURIPS 2022, 2022,
  • [45] Improving Segmentation of Objects with Varying Sizes in Biomedical Images using Instance-wise and Center-of-Instance Segmentation Loss Function
    Rachmadi, Muhammad Febrian
    Poon, Charissa
    Skibbe, Henrik
    MEDICAL IMAGING WITH DEEP LEARNING, VOL 227, 2023, 227 : 286 - 300
  • [46] Accelerating Sparse DNN Models without Hardware-Support via Tile-Wise Sparsity
    Guo, Cong
    Hsuch, Yang
    Leng, Jingwen
    Qiu, Yuxian
    Guan, Yue
    Wang, Zehuan
    Jia, Xiaoying
    Li, Xipeng
    Guo, Minyi
    Zhu, Yuhao
    PROCEEDINGS OF SC20: THE INTERNATIONAL CONFERENCE FOR HIGH PERFORMANCE COMPUTING, NETWORKING, STORAGE AND ANALYSIS (SC20), 2020,
  • [47] Scene Text Detection Based on Multi-Dimensional Feature Fusion with Instance-Wise Loss
    Wu, Qin
    Zhu, Peiwen
    Guo, Guodong
    INTERNATIONAL JOURNAL OF PATTERN RECOGNITION AND ARTIFICIAL INTELLIGENCE, 2023, 37 (05)
  • [48] Instance-wise reaching definition analysis for recursive programs using context-free transductions
    Cohen, A
    Collard, JF
    1998 INTERNATIONAL CONFERENCE ON PARALLEL ARCHITECTURES AND COMPILATION TECHNIQUES, PROCEEDINGS, 1998, : 332 - 339
  • [49] Acorns: A Framework for Accelerating Deep Neural Networks with Input Sparsity
    Dong, Xiao
    Liu, Lei
    Zhao, Peng
    Li, Guangli
    Li, Jiansong
    Wang, Xueying
    Feng, Xiaobing
    2019 28TH INTERNATIONAL CONFERENCE ON PARALLEL ARCHITECTURES AND COMPILATION TECHNIQUES (PACT 2019), 2019, : 178 - 191
  • [50] Step-Wise Deep Learning Models for Solving Routing Problems
    Xin, Liang
    Song, Wen
    Cao, Zhiguang
    Zhang, Jie
    IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, 2021, 17 (07) : 4861 - 4871