• Publications
  • Influence
Deeply-Supervised Nets
TLDR
Our proposed deeply-supervised nets (DSN) method simultaneously minimizes classification error while making the learning process of hidden layers direct and transparent. Expand
  • 1,136
  • 71
  • PDF
GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networks
TLDR
We present a gradient normalization (GradNorm) algorithm that automatically balances training in deep multitask models by dynamically tuning gradient magnitudes. Expand
  • 205
  • 38
  • PDF
Recursive Recurrent Nets with Attention Modeling for OCR in the Wild
TLDR
We present recursive recurrent neural networks with attention modeling (R2AM) for lexicon-free optical character recognition in natural scene images. Expand
  • 231
  • 22
  • PDF
Generalizing Pooling Functions in Convolutional Neural Networks: Mixed, Gated, and Tree
TLDR
We seek to improve deep neural networks by generalizing the pooling operations that play a central role in current architectures. Expand
  • 324
  • 18
  • PDF
Sliced Wasserstein Discrepancy for Unsupervised Domain Adaptation
TLDR
In this work, we connect two distinct concepts for unsupervised domain adaptation: feature distribution alignment between domains by utilizing the task-specific decision boundary and the Wasserstein metric. Expand
  • 99
  • 12
  • PDF
RoomNet: End-to-End Room Layout Estimation
TLDR
This paper focuses on the task of room layout estimation from a monocular RGB image. Expand
  • 82
  • 11
  • PDF
Training Deeper Convolutional Networks with Deep Supervision
TLDR
In order to train deeper networks, we propose to add auxiliary supervision branches after certain intermediate layers during training. Expand
  • 84
  • 9
  • PDF
An O(log n) Parallel Algorithm for Constructing a Spanning Tree on Permutation Graphs
TLDR
We propose an O(log n) time parallel algorithm with O( log n) processors on the EREW PRAM for constructing a spanning tree on an unweighted permutation graph. Expand
  • 11
  • 5
  • PDF
Removing rain and snow in a single image using saturation and visibility features
TLDR
A novel method is proposed based on saturation and visibility features to remove the rain and snow in the photograph. Expand
  • 31
  • 4
Comments on Saeednia's improved scheme for the hill cipher
Abstract In 2000, Saeednia proposed a new scheme to make the Hill cipher secure. The author makes use of permutations of columns and rows of a matrix to get a different key for encrypting eachExpand
  • 23
  • 4