no code implementations • 5 Sep 2023 • David Levin, Gonen Singer
This paper introduces a novel graph-based filter method for automatic feature selection (abbreviated as GB-AFS) for multi-class classification tasks.
no code implementations • 2 Nov 2017 • Barak Sober, Yariv Aizenbud, David Levin
The resulting approximant is shown to be a function defined over a neighborhood of a manifold, approximating the originally sampled manifold.
no code implementations • 22 Jun 2016 • Barak Sober, David Levin
We assume that the data points are located "near" the lower dimensional manifold and suggest a non-linear moving least-squares projection on an approximating $d$-dimensional manifold.
no code implementations • 23 Feb 2016 • Barak Sober, David Levin
This work suggests a new variational approach to the task of computer aided restoration of incomplete characters, residing in a highly noisy document.