Detection of humans and estimation of their 2D poses from a single image are challenging tasks. This is especially true when part of the observation is occluded. However, given a l...
Generative models of pattern individuality attempt to represent the distribution of observed quantitative features, e.g., by learning parameters from a database, and then use such...
Abstract—Segmenting lesions is a vital step in many computerized mass-detection schemes for digital (or digitized) mammograms. We have developed two novel lesion segmentation tec...
We give a general technique to obtain approximation mechanisms that are truthful in expectation. We show that for packing domains, any α-approximation algorithm that also bounds ...
This article extends the concept of compressed sensing to signals that are not sparse in an orthonormal basis but rather in a redundant dictionary. It is shown that a matrix, whic...
Holger Rauhut, Karin Schnass, Pierre Vandergheynst