Currently, I am a PostDoc at the University of Salzburg. My main interest is the connection of machine learning and topological data analysis. In this context, I argue that developing a topological perspective on the phenomenon of learning yields a more formal perspective on the currently rather ``alchemic’’ state of applied machine learning. Probably, my motivation can be explained best in the words of Ali Rahimi at NIPS’17, whose talk was somewhat a point of orientation in the early phase of my PhD.

In particluar, my focus is on the symbiotic combination of deep learning and persistent homology which is a powerful tool to derive topological properties of data by means of algebra. Beneath my research, I regularly serve as a reviewer for machine learning conferences, e.g., ICML, NeurIPS.

brief resumee

• Oct. 2007 - Feb. 2014 : Study of Mathematics at the University of Salzburg

• Mar. 2014 - Sep. 2015 : Software engineer, COPA-DATA group

• Oct. 2015 - Jan. 2020 : PhD in Computer Science at the University of Salzburg

• Feb. 2020 - now : PostDoc position at the University of Salzburg

For the official 30 seconds resumee click here.

arXiv preprints

• Graph Filtration Learning
• Topologically Densified Distributions

publications

• Learning Representations of Persistence Barcodes, JMLR’19
• Connectivity-optimized representation learning via persistent homology, ICML’19
• Deep Learning with Topological Signatures, NIPS’17
• Constructing Shape Spaces from a Topological Perspective, IPMI’17
• Simple domain adaptation for cross-dataset analyses of brain MRI data, ISBI’17