Matthäus Kleindessner
University of Tübingen
Sand 14, 72076 Tübingen, Germany
Room: B220
Phone: +49 7071 29-77173
E-mail: matthaeus.kleindessner@uni-tuebingen.de
Sand 14, 72076 Tübingen, Germany
Room: B220
Phone: +49 7071 29-77173
E-mail: matthaeus.kleindessner@uni-tuebingen.de
From 2006 to 2012 I have studied Technical Mathematics at the Vienna University of Technology. Since March 2013 I am a PhD student in the group of Ulrike von Luxburg, which was affiliated to the University of Hamburg until October 2015.
My work deals with ordinal data analysis. Ordinal data (also known as ordinal distance information) refers to only binary answers to distance comparisons between objects in some data set rather than actual numerical distance measurements. There are many situations in which such data is encountered (e.g., in crowdsourcing scenarios). The aim of my work is to theoretically understand ordinal data and to provide algorithms for machine learning problems based on it.
Related publications:
- M. Kleindessner, U. von Luxburg. Lens depth function and k-relative neighborhood graph: versatile tools for ordinal data analysis. Journal of Machine Learning Research 18(58):1-52, 2017. pdf, data, large version of the embedding of the car data set
- M. Kleindessner, U. von Luxburg. Kernel functions based on triplet similarity comparisons. Preprint to be found on arXiv, 2016. pdf
- M. Kleindessner, U. von Luxburg. Dimensionality estimation without distances. International Conference on Artificial Intelligence and Statistics (AISTATS), 2015. pdf , poster
- M. Kleindessner, U. von Luxburg. Uniqueness of ordinal embedding. Conference on Learning Theory (COLT), 2014. pdf