• Publications
  • Influence
PD-Sparse : A Primal and Dual Sparse Approach to Extreme Multiclass and Multilabel Classification
TLDR
We show that a margin-maximizing loss with l1 penalty, in case of Extreme Classification, yields extremely sparse solution both in primal and in dual without sacrificing the expressive power of predictor. Expand
  • 119
  • 25
  • PDF
Recovery Guarantees for One-hidden-layer Neural Networks
TLDR
We distill some properties of activation functions that lead to $\mathit{local~strong~convexity}$ in the neighborhood of the ground-truth parameters for the 1NN squared-loss objective. Expand
  • 216
  • 19
  • PDF
The molecular basis for gray and white matter contrast in phase imaging
TLDR
This study demonstrates that the water resonance frequency is directly shifted by water-macromolecule exchange processes (0.040 ppm/mM for bovine serum albumin) and might be a major source of contribution to phase image contrast. Expand
  • 123
  • 11
  • PDF
Efficient Matrix Sensing Using Rank-1 Gaussian Measurements
TLDR
In this paper, we propose simple rank-one Gaussian measurement operators for matrix sensing that are significantly less expensive in terms of memory and computation for both encoding and decoding. Expand
  • 34
  • 8
  • PDF
Brain Metabolite Abnormalities in the White Matter of Elderly Schizophrenic Subjects: Implication for Glial Dysfunction
BACKGROUND Abnormalities in the white matter of the brain may occur in individuals with schizophrenia as well as with normal aging. Therefore, elderly schizophrenic patients may suffer furtherExpand
  • 94
  • 7
MR-Encephalography: Fast multi-channel monitoring of brain physiology with magnetic resonance
TLDR
A new approach to measure activation-related changes in the brain by magnetic resonance is described offering high temporal resolution of 10-100 measurements per second, which allows to directly measure the stimulus-dependent hemodynamic response as well as ECG- and breathing-related signal fluctuations. Expand
  • 75
  • 7
Learning Non-overlapping Convolutional Neural Networks with Multiple Kernels
TLDR
In this paper, we consider parameter recovery for non-overlapping convolutional neural networks (CNNs) with multiple kernels. Expand
  • 57
  • 7
  • PDF
Efficient and Non-Convex Coordinate Descent for Symmetric Nonnegative Matrix Factorization
TLDR
In this paper, we propose simple and very efficient coordinate descent schemes, which solve symNMF by solving a series of fourth-order univariate subproblems exactly. Expand
  • 36
  • 6
  • PDF
Levels and Trends in Child Mortality : Report 2018
Despite progress over the past quarter-century, millions of newborns, children and young adolescents die every year, mostly of preventable or treatable causes such as infectious diseasesand injuries.Expand
  • 59
  • 5
Mixed Linear Regression with Multiple Components
TLDR
In this paper, we study the mixed linear regression (MLR) problem, where the goal is to recover multiple underlying linear models from their unlabeled linear measurements. Expand
  • 31
  • 5
  • PDF