2020/10/15 by Martin Keller-Ressel, Keller-Ressel, Martin · 1 citation
Computer Science · Mathematics · #62J02 #68T07 #FOS: Computer and information sciences #G.3 #I.5 #Machine Learning (cs.LG) #Machine Learning (stat.ML) #acm:62J02 #acm:68T07 #cs.LG #msc:62J02 #msc:68T07 #stat.ML
paper · pdf · doi:10.48550/arxiv.2010.07744
6 pages
arxiv created 2020/10/15 · arxiv updated 2020/10/16
We introduce Hyperbolic Prototype Learning, a type of supervised learning, where class labels are represented by ideal points (points at infinity) in hyperbolic space. Learning is achieved by minimizing the 'penalized Busemann loss', a new loss function based on the Busemann function of hyperbolic geometry. We discuss several theoretical features of this setup. In particular, Hyperbolic Prototype Learning becomes equivalent to logistic regression in the one-dimensional case.