Advertisement

User Loss A Forced-Choice-Inspired Approach to Train Neural Networks Directly by User Interaction

  • Shahab Zarei
  • Bernhard Stimpel
  • Christopher Syben
  • Andreas MaierEmail author
Conference paper
Part of the Informatik aktuell book series (INFORMAT)

Zusammenfassung

In this paper, we investigate whether is it possible to train a neural network directly from user inputs. We consider this approach to be highly relevant for applications in which the point of optimality is not well-defined and user-dependent. Our application is medical image denoising which is essential in fluoroscopy imaging. In this field every user, i.e. physician, has a different flavor and image quality needs to be tailored towards each individual. To address this important problem, we propose to construct a loss function derived from a forced-choice experiment. In order to make the learning problem feasible, we operate in the domain of precision learning, i.e., we inspire the network architecture by traditional signal processing methods in order to reduce the number of trainable parameters. The algorithm that was used for this is a Laplacian pyramid with only six trainable parameters. In the experimental results, we demonstrate that two image experts who prefer different filter characteristics between sharpness and de-noising can be created using our approach. Also models trained for a specific user perform best on this users test data. This approach opens the way towards implementation of direct user feedback in deep learning and is applicable for a wide range of application.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Literatur

  1. 1.
    LeCun Y, Bengio Y, Hinton G. Deep learning. Nat. 2015;521(7553):436.CrossRefGoogle Scholar
  2. 2.
    Maier AK, Schebesch F, Syben C, et al. Precision learning: towards use of known operators in neural networks. CoRR. 2017;abs/1712.00374. Available from: http://arxiv.org/abs/1712.00374.
  3. 3.
    Syben C, Stimpel B, Breininger K, et al. Precision learning: reconstruction filter kernel discretization. Procs Fifth Int Conf Image Form X-Ray Comput Tomogr. 2018; p. 386–390.Google Scholar
  4. 4.
    Fu W, Breininger K, Schaffert R, et al. Frangi-net: a neural network approach to vessel segmentation. BVM 2018. 2018; p. 341–346.CrossRefGoogle Scholar
  5. 5.
    Rajashekar U, Simoncelli EP. Multiscale denoising of photographic images. Essent Guide Image Process. 2009; p. 241–261.Google Scholar
  6. 6.
    Rockafellar RT. Convex Analysis. Princeton landmarks in mathematics and physics. Princeton University Press; 1970. Available from: https://books.google.de/books?id=1TiOka9bx3sC.
  7. 7.
    Rumelhart DE, Hinton GE, Williams RJ. Learning representations by backpropagating errors. Nat. 1986;323(6088):533.CrossRefGoogle Scholar
  8. 8.
    Bishop CM. Pattern Recognition and Machine Learning (Information Science and Statistics). Berlin, Heidelberg: Springer-Verlag; 2006.Google Scholar
  9. 9.
    Tomasi C, Manduchi R; IEEE. Bilateral filtering for gray and color images. Proc ICCV. 1998; p. 839–846.Google Scholar
  10. 10.
    Petschnigg G, Szeliski R, Agrawala M, et al.; ACM. Digital photography with ash and no-ash image pairs. ACM Trans graph (TOG). 2004;23(3):664–672.CrossRefGoogle Scholar
  11. 11.
    Luisier F, Blu T, Unser M. A new SURE approach to image denoising: interscale orthonormal wavelet thresholding. IEEE Trans Image Process. 2007;16(3):593–606.MathSciNetCrossRefGoogle Scholar
  12. 12.
    Motwani MC, Gadiya MC, Motwani RC, et al. Survey of image denoising techniques. Procs GSPX. 2004; p. 27–30.Google Scholar

Copyright information

© Springer Fachmedien Wiesbaden GmbH, ein Teil von Springer Nature 2019

Authors and Affiliations

  • Shahab Zarei
    • 1
  • Bernhard Stimpel
    • 1
  • Christopher Syben
    • 1
  • Andreas Maier
    • 1
    Email author
  1. 1.Pattern Recognition Lab, Department of Computer ScienceFriedrich-Alexander-Universität Erlangen-NürnbergErlangenDeutschland

Personalised recommendations