• Publications
  • Influence
Deep High-Resolution Representation Learning for Visual Recognition
TLDR
We present a novel architecture, namely High-Resolution Network (HRNet), which is able to maintain high-resolution representations through the whole process. Expand
  • 157
  • 38
  • PDF
Discrimination-aware Channel Pruning for Deep Neural Networks
TLDR
We investigate a simple-yet-effective method, called discrimination-aware channel pruning, to choose those channels that really contribute to discriminative power. Expand
  • 206
  • 21
  • PDF
Learning Sparse SVM for Feature Selection on Very High Dimensional Datasets
TLDR
In this paper, by introducing a 0-1 control variable to each input feature, l0-norm Sparse SVM is converted to a mixed integer programming (MIP) problem. Expand
  • 172
  • 19
  • PDF
Heterogeneous Domain Adaptation for Multiple Classes
TLDR
In this paper, we present an efficient multi-class heterogeneous domain adaptation method, where data from source and target domains are represented by heterogeneous features of different dimensions. Expand
  • 100
  • 15
  • PDF
Domain-Symmetric Networks for Adversarial Domain Adaptation
TLDR
We propose a novel domain adaptation method based on a symmetric design of source and target task classifiers, based on which we also construct an additional classifier that shares with them its layer neurons. Expand
  • 67
  • 15
  • PDF
Graph Convolutional Networks for Temporal Action Localization
TLDR
We propose to exploit the proposal-proposal relations using GraphConvolutional Networks (GCNs) to model the relations among different proposals and learn powerful representations for the action classification and localization. Expand
  • 77
  • 12
  • PDF
Gene selection using hybrid particle swarm optimization and genetic algorithm
TLDR
A hybrid Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) method is used for gene selection, and Support Vector Machine (SVM) is adopted as the classifier. Expand
  • 116
  • 10
  • PDF
Weighted Block-Sparse Low Rank Representation for Face Clustering in Videos
TLDR
We propose a new method called weighted block-sparse low rank representation (WBSLRR) which considers the available prior knowledge while learning a low rank data representation, and also develop a simple but effective approach to obtain the clustering result of faces. Expand
  • 78
  • 10
  • PDF
Visual Grounding via Accumulated Attention
TLDR
Visual Grounding (VG) aims to locate the most relevant object or region in an image, based on a natural language query. Expand
  • 74
  • 9
  • PDF
Fast Algorithms for Linear and Kernel SVM+
TLDR
The SVM+ approach has shown excellent performance in visual recognition tasks for exploiting privileged information in the training data, and not available during the test stage. Expand
  • 27
  • 9
  • PDF