Abstract--This paper presents a novel and domainindependent approach for graph-based structure learning. The approach is based on solving the Maximum Common SubgraphIsomorphism pro...
We consider the problem of learning classifiers in structured domains, where some objects have a subset of features that are inherently absent due to complex relationships between...
Gal Chechik, Geremy Heitz, Gal Elidan, Pieter Abbe...
A geometric approach is introduced to explain phenomena that can arise with Luce's choice axiom; e.g., differences occur when determining the likelihood of a ranking by start...
We present a novel method for unsupervised classification, including the discovery of a new category and precise object and part localization. Given a set of unlabelled images, som...
Leonid Karlinsky, Michael Dinerstein, Dan Levi, Sh...
We study geometric reconstruction problems in one-dimensional retina vision. In such problems, the scene is modeled as a 2D plane, and the camera sensor produces 1D images of the s...
Olof Enqvist, Fredrik Kahl, Carl Olsson, Kalle &Ar...