Convolutional neural networks analysed via inverse problem theory and sparse representations

Convolutional neural networks analysed via inverse problem theory and sparse representations

For access to this article, please select a purchase option:

Buy eFirst article PDF
(plus tax if applicable)
Buy Knowledge Pack
10 articles for £75.00
(plus taxes if applicable)

IET members benefit from discounts to all IET publications and free access to E&T Magazine. If you are an IET member, log in to your account and the discounts will automatically be applied.

Learn more about IET membership 

Recommend Title Publication to library

You must fill out fields marked with: *

Librarian details
Your details
Why are you recommending this title?
Select reason:
IET Signal Processing — Recommend this title to your library

Thank you

Your recommendation has been sent to your librarian.

Inverse problems in imaging such as denoising, deblurring, superresolution have been addressed for many decades. In recent years, convolutional neural networks (CNNs) have been widely used for many inverse problem areas. Although their indisputable success, CNNs are not mathematically validated as to how and what they learn. In this study, the authors prove that during training, CNN elements solve for inverse problems which are optimum solutions stored as CNN neuron filters. They discuss the necessity of mutual coherence between CNN layer elements in order for a network to converge to the optimum solution. They prove that required mutual coherence can be provided by the usage of residual learning and skip connections. They have set rules over training sets and depth of networks for better convergence, i.e. performance. They have experimentally validated theoretical assertions.


    1. 1)
      • 1. LeCun, Y., Bengio, Y., Hinton, G.E.: ‘Deep learning’, Nature, 2015, 521, (7553), pp. 436444.
    2. 2)
      • 2. McCann, M.T., Jin, K.H., Unser, M.: ‘Convolutional neural networks for inverse problems in imaging: a review’, IEEE Signal Process. Mag., 2017, 34, (6), pp. 8595.
    3. 3)
      • 3. Lucas, A., Iliadis, M., Molina, R., et al: ‘Using deep neural networks for inverse problems in imaging: beyond analytical methods’, IEEE Signal Process. Mag., 2018, 35, (1), pp. 2036.
    4. 4)
      • 4. Gregor, K., LeCun, Y.: ‘Learning fast approximations of sparse coding’. Proc. 27th Int. Conf. Machine Learning (ICML-10), Haifa, 2010, pp. 399406.
    5. 5)
      • 5. Yang, Y., Sun, J., Li, H., et al: ‘Deep admm-net for compressive sensing MRI’. Advances in Neural Information Processing Systems 29: Annual Conf. on Neural Information Processing Systems 2016, Barcelona, Spain, 5–10 December 2016, pp. 1018.
    6. 6)
      • 6. Dong, C., Loy, C.C., He, K., et al: ‘Learning a deep convolutional network for image super-resolution’. European Conf. on Computer Vision (ECCV) (4), Zurich, September 2014 (LNCS, 8692), pp. 184199.
    7. 7)
      • 7. Bengio, Y., Courville, A.C., Vincent, P.: ‘Representation learning: A review and new perspectives’, IEEE Trans. Pattern Anal. Mach. Intell., 2013, 35, (8), pp. 17981828.
    8. 8)
      • 8. Papyan, V., Romano, Y., Elad, M.: ‘Convolutional neural networks analyzed via convolutional sparse coding’, CoRR, 2016, 18, pp. 83:183:52.
    9. 9)
      • 9. Mao, X., Shen, C., Yang, Y.: ‘Image restoration using very deep convolutional encoder-decoder networks with symmetric skip connections’. Advances in Neural Information Processing Systems 29: Annual Conf. Neural Information Processing Systems, Barcelona, 2016, pp. 28022810.
    10. 10)
      • 10. Kim, J., Lee, J.K., Lee, K.M.: ‘Accurate image super-resolution using very deep convolutional networks’. Computer Vision and Pattern Recognition (CVPR), Las Vegas, 2016, pp. 16461654.
    11. 11)
      • 11. Yang, J., Wright, J., Huang, T.S., et al: ‘Image super-resolution as sparse representation of raw image patches’. 2008 IEEE Computer Society Conf. Computer Vision and Pattern Recognition (CVPR 2008), Anchorage, June 2008.
    12. 12)
      • 12. Yang, J., Wright, J., Huang, T.S., et al: ‘Image super-resolution via sparse representation’, Trans. Image Process., 2010, 19, (11), pp. 28612873.
    13. 13)
      • 13. Timofte, R., Smet, V.D., Gool, L.J.V.: ‘Anchored neighborhood regression for fast example-based super-resolution’. Int. Conf. Computer Vision (ICCV), Portland, 2013, pp. 19201927.
    14. 14)
      • 14. Timofte, R., Smet, V.D., Gool, L.J.V.: ‘A+: adjusted anchored neighborhood regression for fast super-resolution’. ACCV (4), Singapore, 2014 (LNCS, 9006), pp. 111126.
    15. 15)
      • 15. Romano, Y., Isidoro, J., Milanfar, P.: ‘RAISR: rapid and accurate image super resolution’, IEEE Trans. Comput. Imag., 2017, 3, (1), pp. 110125.
    16. 16)
      • 16. Chen, Y., Pock, T.: ‘Trainable nonlinear reaction diffusion: a flexible framework for fast and effective image restoration’, IEEE Trans. Pattern Anal. Mach. Intell., 2017, 39, (6), pp. 12561272.
    17. 17)
      • 17. Aharon, M., Elad, M., Bruckstein, A.: ‘Ksvd: an algorithm for designing overcomplete dictionaries for sparse representation’, Trans. Signal Process., 2006, 54, (11), pp. 43114322.
    18. 18)
      • 18. Combettes, P.L., Wajs, V.R.: ‘Signal recovery by proximal forward-backward splitting’, Multiscale. Model. Simul., 2005, 4, (4), pp. 11681200.
    19. 19)
      • 19. Bronstein, A.M., Sprechmann, P., Sapiro, G.: ‘Learning efficient structured sparse models’. Proc. 29th Int. Conf. Machine Learning (ICML), Edinburgh, 2012.
    20. 20)
      • 20. Schuler, C.J., Hirsch, M., Harmeling, S., et al: ‘Learning to deblur’, IEEE Trans. Pattern Anal. Mach. Intell., 2016, 38, (7), pp. 14391451.
    21. 21)
      • 21. Zhang, K., Zuo, W., Chen, Y., et al: ‘Beyond a Gaussian denoiser: residual learning of deep CNN for image denoising’, IEEE Trans. Image Process., 2017, 26, (7), pp. 31423155.
    22. 22)
      • 22. Gabay, D., Mercier, B.: ‘A dual algorithm for the solution of nonlinear variational problems via finite element approximation’, Comput. Math. Appl., 1976, 2, (1), pp. 1740.
    23. 23)
      • 23. Maas, A.L., Hannun, A.Y., Ng, A.Y.: ‘Rectifier nonlinearities improve neural network acoustic models’. ICML Workshop on Deep Learning for Audio, Speech and Language Processing, Atlanta, 2013.
    24. 24)
      • 24. Glorot, X., Bordes, A., Bengio, Y.Deep sparse rectifier neural networks’. Proc. Artificial Intelligence and Statistics (AISTATS), Fort Lauderdale, April 2011, vol. 15,, pp. 315323.
    25. 25)
      • 25. Bruna, J., Sprechmann, P., LeCun, Y.: ‘Super-resolution with deep convolutional sufficient statistics’, CoRR, 2015, available at
    26. 26)
      • 26. Shi, W., Caballero, J., Huszar, F., et al: ‘Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network’. Computer Vision and Pattern Recognition (CVPR), Las Vegas, 2016, pp. 18741883.
    27. 27)
      • 27. Daubechies, I., Defrise, M., De Mol, C.: ‘An iterative thresholding algorithm for linear inverse problems with a sparsity constraint’, Commun. Pure Appl. Math., 2004, 57, (11), pp. 14131457.
    28. 28)
      • 28. Mallat, S.: ‘Understanding deep convolutional networks’, CoRR, 2016, available at:
    29. 29)
      • 29. Papyan, V., Sulam, J., Elad, M.: ‘Working locally thinking globally: theoretical guarantees for convolutional sparse coding’, IEEE Trans. Signal Process., 2017, 65, (21), pp. 56875701.

Related content

This is a required field
Please enter a valid email address