eduzhai > Applied Sciences > Engineering >

On the unreasonable effectiveness of CNNs

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 4 pages

Abstract: Deep learning methods using convolutional neural networks (CNN) have beensuccessfully applied to virtually all imaging problems, and particularly inimage reconstruction tasks with ill-posed and complicated imaging models. In anattempt to put upper bounds on the capability of baseline CNNs for solvingimage-to-image problems we applied a widely used standard off-the-shelf networkarchitecture (U-Net) to the "inverse problem " of XOR decryption from noisy dataand show acceptable results.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×