• Publications
  • Influence
SPIDER: Near-Optimal Non-Convex Optimization via Stochastic Path Integrated Differential Estimator
TLDR
In this paper, we propose a new technique named Stochastic Path-Integrated Differential EstimatoR (SPIDER), which can be used to track many deterministic quantities of interest with significantly reduced computational cost. Expand
  • 207
  • 67
  • PDF
Deep Subspace Clustering Networks
TLDR
We present a novel deep neural network architecture to learn (in an unsupervised manner) an explicit non-linear mapping of the data that is well-adapted to subspace clustering. Expand
  • 179
  • 47
  • PDF
Deep Learning Based Feature Selection for Remote Sensing Scene Classification
TLDR
A novel deep-learning-based feature-selection method has been proposed, which formulates the featureselection problem as a feature reconstruction problem in a DBN. Expand
  • 335
  • 38
Efficient mini-batch training for stochastic optimization
TLDR
Stochastic gradient descent (SGD) is a popular technique for large-scale optimization problems in machine learning. Expand
  • 426
  • 26
  • PDF
Learning to Remember Translation History with a Continuous Cache
TLDR
We propose to augment NMT models with a very light-weight cache-like memory network, which stores recent hidden representations as translation history. Expand
  • 92
  • 19
  • PDF
Stochastic Optimization with Importance Sampling for Regularized Loss Minimization
TLDR
Uniform sampling of training data has been commonly used in traditional stochastic optimization algorithms such as Proximal Stochastic Mirror Descent (prox-SMD) and prox-SDCA. Expand
  • 195
  • 18
  • PDF
Stochastic Optimization with Importance Sampling
TLDR
In this paper we study stochastic optimization with importance sampling, which improves the convergence rate for prox-SGD and prox-SDCA. Expand
  • 76
  • 15
  • PDF
ARGs-OAP: online analysis pipeline for antibiotic resistance genes detection from metagenomic data using an integrated structured ARG-database
TLDR
We developed an online pipeline called ARGs-OAP for fast annotation and classification of ARG-like sequences from metagenomic data from environmental samples. Expand
  • 114
  • 12
Super-Identity Convolutional Neural Network for Face Hallucination
TLDR
This paper proposes Super-Identity Convolutional Neural Network (SICNN) to recover identity information for generating faces closed to the real identity. Expand
  • 58
  • 12
  • PDF
A multistage, multimethod approach for automatic detection and classification of epileptiform EEG
TLDR
An efficient system for detection of epileptic activity in ambulatory electroencephalogram (EEG) that combines multiple signal-processing methods in a multistage scheme, integrating adaptive filtering, wavelet transform, an artificial neural network, and expert system. Expand
  • 120
  • 10
  • PDF
...
1
2
3
4
5
...