• Publications
  • Influence
Robust Large Margin Deep Neural Networks
TLDR
The generalization error of deep neural networks via their classification margin is studied in this paper. Expand
  • 166
  • 26
  • PDF
MeshCNN: a network with an edge
TLDR
We utilize the unique properties of the mesh for direct analysis of 3D shapes using MeshCNN, a convolutional neural network designed specifically for triangular meshes. Expand
  • 94
  • 15
  • PDF
Learned Convolutional Sparse Coding
TLDR
We propose a convolutional recurrent sparse auto-encoder model that learns a task-driven sparse Convolutional Dictionary and produces an approximate sparse code over the learned dictionary. Expand
  • 43
  • 12
  • PDF
Improving DNN Robustness to Adversarial Attacks using Jacobian Regularization
TLDR
We apply the Frobenius norm of the Jacobian of the network, which is applied as post-processing, after regular training has finished. Expand
  • 78
  • 9
  • PDF
ASAP: Architecture Search, Anneal and Prune
TLDR
In this paper, we propose a differentiable search space that allows the annealing of architecture weights, while gradually pruning inferior operations. Expand
  • 41
  • 9
  • PDF
Poisson Inverse Problems by the Plug-and-Play scheme
TLDR
The Anscombe transform offers an approximate conversion of a Poisson random variable into a Gaussian one, and becomes handy in various inverse problems with Poisson noise contamination. Expand
  • 71
  • 8
  • PDF
Image Restoration by Iterative Denoising and Backward Projections
  • Tom Tirer, Raja Giryes
  • Computer Science, Medicine
  • IEEE Transactions on Image Processing
  • 18 October 2017
TLDR
We propose an alternative method for solving linear inverse problems using off-the-shelf denoisers, which requires less parameter tuning than the P&P approach for image inpainting and deblurring. Expand
  • 54
  • 8
  • PDF
Delta-encoder: an effective sample synthesis method for few-shot object recognition
TLDR
We propose a simple yet effective method for few-shot (and one-shot) object-recognition, based on a modified auto-encoder that learns to synthesize new samples for an unseen category just by seeing few examples from it. Expand
  • 95
  • 7
  • PDF
The Projected GSURE for Automatic Parameter Tuning in Iterative Shrinkage Methods
TLDR
We use the projected Generalized Stein Unbiased Risk Estimator (GSURE) for determining the threshold value lambda in iterative shrinkage methods for image deblurring and image zooming. Expand
  • 118
  • 7
  • PDF
Sparsity-Based Poisson Denoising With Dictionary Learning
TLDR
The problem of Poisson denoising appears in various imaging applications, such as low-light photography, medical imaging, and microscopy. Expand
  • 81
  • 7
  • PDF