Bastian Rieck is a postdoctoral researcher in the Machine Learning and Computational Biology Lab of Prof. Dr. Karsten Borgwardt at ETH Zurich. His main research interests are algorithms for graph classification and time series analysis. Bastian received his M.Sc. degree in mathematics, as well as his Ph.D., from Heidelberg University in Germany.

Title of the talk

Perspectives in Persistent Homology

Abstract

This discusses future perspectives for research in persistent homology with a specific focus towards machine learning. I will first outline and summarise research from the decades, pointing out connections to machine learning. This is followed by a discussion of potential future applications as well as future research directions. Finally, I shall conclude this talk by giving suggestions for increasing the visibility of persistent homology (and, more generically, of TDA) in the hopes of providing stimulus to the nascent---but highly intriguing---field of topological machine learning.

Marcello Paris comes from a Ph.D. in Mathematics and spent many years in market finance and computational IT, mainly for applications to derivatives modelling and risk measurement. His current research interests are topological methods in data analysis, reinforcement learning and smart contracts.

Title of the talk

Homology-based learning: examples from time series and point clouds

Abstract

The talk is about persistent homology and its applications to entropy measures and optimization problems. I’ll discuss the category-theory approach, a use of non-persistent features in barcodes, examples from (chaotic) dynamical systems and a method to let algorithms learn interesting subsets of point clouds.

Jan has a background in genetics and genomics, and performed his doctoral research at the University of Wageningen (Netherlands) on the Chicken Genome Sequencing Project. He then moved to Scotland to work as a postdoctoral researcher at the Roslin Institute on the Cow Genome Sequencing Project. Next, he continued his research at the Wellcome Trust Sanger Institute near Cambridge (UK) focusing on structural variation in the human and other primate genomes. At his return to Belgium at the KU Leuven in 2010 (as assistent and later associate professor), he shifted focus to data visualization and visual analytics, with the aim of finding interesting questions in large datasets (big data). His main research topics revolve around visual design, interaction design, and (human and computational) scalability. Since 2019, he is professor at Hasselt University where he continues his visual analytics work and helps build a new Data Science Institute.

Jan has been on the organising committees of several conferences (including BioVis and Beyond The Genome), and has chaired visualization-related sessions at conferences including VIZBI, the Bioinformatics Open Source Conference BOSC and EuroVis/VMLS. He is also Associate Editor for the BioMedCentral Thematic Series on Biological Data Visualization, and academic editor for PLoS One. He was founding member of the Young Academy – Royal Flemish Academy of Belgium for Sciences and the Arts.

Title of the talk

Topological Data Analysis in Visual Analytics

Abstract

In this talk, I'll talk about the reasons to use visual analytics in data science, and what role topological data analysis can play therein. I'll cover some of the issues with current data analysis methods as well as the strengths and weaknesses of data visualisation, and will provide some examples of complex datasets viewed from a topological perspective.