DSConv: Efficient Convolution Operator

被引:30
|
作者
do Nascimento, Marcelo Gennari [1 ]
Fawcett, Roger [2 ]
Prisacariu, Victor Adrian [1 ]
机构
[1] Univ Oxford, Act Vis Lab, Oxford, England
[2] Intel Corp, Santa Clara, CA 95051 USA
关键词
D O I
10.1109/ICCV.2019.00525
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Quantization is a popular way of increasing the speed and lowering the memory usage of Convolution Neural Networks (CNNs). When labelled training data is available, network weights and activations have successfully been quantized down to 1-bit. The same cannot be said about the scenario when labelled training data is not available, e.g. when quantizing a pre-trained model, where current approaches show, at best, no loss of accuracy at 8-bit quantizations. We introduce DSConv, a flexible quantized convolution operator that replaces single-precision operations with their far less expensive integer counterparts, while maintaining the probability distributions over both the kernel weights and the outputs. We test our model as a plug-and-play replacement for standard convolution on most popular neural network architectures, ResNet, DenseNet, GoogLeNet, AlexNet and VGG-Net and demonstrate state-of-the-art results, with less than 1% loss of accuracy, without retraining, using only 4-bit quantization. We also show how a distillation-based adaptation stage with unlabelled data can improve results even further.
引用
收藏
页码:5147 / 5156
页数:10
相关论文
共 50 条
  • [1] Optimized separable convolution: Yet another efficient convolution operator
    Wei, Tao
    Tian, Yonghong
    Wang, Yaowei
    Liang, Yun
    Chen, Chang Wen
    [J]. AI OPEN, 2022, 3 : 162 - 171
  • [2] SpiralNet plus plus : A Fast and Highly Efficient Mesh Convolution Operator
    Gong, Shunwang
    Chen, Lei
    Bronstein, Michael
    Zafeiriou, Stefanos
    [J]. 2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION WORKSHOPS (ICCVW), 2019, : 4141 - 4148
  • [3] On a Generalized Convolution Operator
    Sharma, Poonam
    Raina, Ravinder Krishna
    Sokol, Janusz
    [J]. SYMMETRY-BASEL, 2021, 13 (11):
  • [4] Entropy of a convolution operator
    Urbanski, A
    [J]. OPEN SYSTEMS & INFORMATION DYNAMICS, 2004, 11 (01): : 79 - 85
  • [5] A CONVOLUTION INTEGRAL OPERATOR
    STEPANOV, VD
    [J]. SIBERIAN MATHEMATICAL JOURNAL, 1982, 23 (02) : 243 - 255
  • [6] RANGE OF AN OPERATOR OF CONVOLUTION TYPE
    EPIFANOV, OV
    [J]. MATHEMATICAL NOTES, 1978, 24 (1-2) : 538 - 543
  • [7] Spectrum of a convolution operator with potential
    Borisov, D. I.
    Zhizhina, E. A.
    Piatnitski, A. L.
    [J]. RUSSIAN MATHEMATICAL SURVEYS, 2022, 77 (03) : 546 - 548
  • [8] On the spectrum of convolution operator with a potential
    Borisov, D. I.
    Piatnitski, A. L.
    Zhizhina, E. A.
    [J]. JOURNAL OF MATHEMATICAL ANALYSIS AND APPLICATIONS, 2023, 517 (01)
  • [9] On the convolution operator in Morrey spaces
    Nursultanov, Erlan D.
    Suragan, Durvudkhan
    [J]. JOURNAL OF MATHEMATICAL ANALYSIS AND APPLICATIONS, 2022, 515 (01)
  • [10] CONVOLUTION OPERATOR ON A FINITE INTERVAL
    LJUBARSKII, JI
    [J]. MATHEMATICS OF THE USSR-IZVESTIYA, 1977, 11 (03): : 583 - 611