Diagnosing Convolutional Neural Networks using their Spectral Response

被引:0
|
作者
Stamatescu, Victor [1 ]
McDonnell, Mark D. [2 ]
机构
[1] Def Sci & Technol Grp, Edinburgh, SA 5111, Australia
[2] Univ South Australia, Sch Informat Technol & Math Sci, Computat Learning Syst Lab Cls Lab Org, Mawson Lakes, SA 5095, Australia
关键词
D O I
暂无
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Convolutional Neural Networks (CNNs) are a class of artificial neural networks whose computational blocks use convolution, together with other linear and non-linear operations, to perform classification or regression. This paper explores the spectral response of CNNs and its potential use in diagnosing problems with their training. We measure the gain of CNNs trained for image classification on ImageNet and observe that the best models are also the most sensitive to perturbations of their input. Further, we perform experiments on MNIST and CIFAR-10 to find that the gain rises as the network learns and then saturates as the network converges. Moreover, we find that strong gain fluctuations can point to overfitting and learning problems caused by a poor choice of learning rate. We argue that the gain of CNNs can act as a diagnostic tool and potential replacement for the validation loss when hold-out validation data are not available.
引用
收藏
页码:603 / 610
页数:8
相关论文
共 50 条
  • [1] Spatially resolved spectral unmixing using convolutional neural networks
    Anastasiadis, Johannes
    Leon, Fernando Puente
    TM-TECHNISCHES MESSEN, 2019, 86 : S122 - S126
  • [2] Spatially resolved spectral unmixing using convolutional neural networks
    Anastasiadis, Johannes
    Puente León, Fernando
    Technisches Messen, 2019, 86 (s1):
  • [3] Spectral Representations for Convolutional Neural Networks
    Rippel, Oren
    Snoek, Jasper
    Adams, Ryan P.
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 28 (NIPS 2015), 2015, 28
  • [4] Opportunistic use of Spectral Holes in Karachi using Convolutional Neural Networks
    Shaikh, Aamir Zeb
    Naqvi, Shabbar
    Ali, Minaal
    Iqbal, Yamna
    Rahim, Abdul
    Khadim, Saima
    Altaf, Talat
    INTERNATIONAL JOURNAL OF ADVANCED COMPUTER SCIENCE AND APPLICATIONS, 2020, 11 (02) : 776 - 780
  • [5] Robust multimaterial decomposition of spectral CT using convolutional neural networks
    Chen, Zhengyang
    Li, Liang
    OPTICAL ENGINEERING, 2019, 58 (01)
  • [6] Epileptic Seizure Detection using Spectral Transformation and Convolutional Neural Networks
    Cleatus T.S.
    Thungamani M.
    Journal of The Institution of Engineers (India): Series B, 2022, 103 (04) : 1115 - 1125
  • [7] Opportunistic use of spectral holes in Karachi using convolutional neural networks
    Shaikh A.Z.
    Naqvi S.
    Ali M.
    Iqbal Y.
    Rahim A.
    Khadim S.
    Altaf T.
    1600, Science and Information Organization : 776 - 780
  • [8] Transferability of spectral graph convolutional neural networks
    Levie, Ron
    Huang, Wei
    Bucci, Lorenzo
    Bronstein, Michael
    Kutyniok, Gitta
    Journal of Machine Learning Research, 2021, 22
  • [9] DIRECTION FINDING USING CONVOLUTIONAL NEURAL NETWORKS and CONVOLUTIONAL RECURRENT NEURAL NETWORKS
    Uckun, Fehmi Ayberk
    Ozer, Hakan
    Nurbas, Ekin
    Onat, Emrah
    2020 28TH SIGNAL PROCESSING AND COMMUNICATIONS APPLICATIONS CONFERENCE (SIU), 2020,
  • [10] Convolutional neural networks used for random structure SPP gratings spectral response prediction
    Qu, Tianle
    Zhu, Liping
    An, Zhenghua
    OPTICS LETTERS, 2023, 48 (02) : 448 - 451