Professor Jilles Vreeken: Learning Actionable Insights from Scientific Data

Professor Jilles Vreeken will give a talk on Learning Actionable Insights from Scientific Data.

Abstract
Many problems in science can be phrased as `under what conditions does something of interest happen?'. In cancer research, we have therapies that work well for some patients but are very nasty for others. Wouldn't it be great if we can automatically find those subgroups of people with exceptional survival characteristics? In materials science, we are after highly effective photo-voltaic panels. Wouldn't it be great if we can automatically characterize the atomistic properties of the most effective molecules in an automated manner? In pharmaceutics, we are interested in the causes of anti-microbial resistance. Wouldn't it be great if we can automatically find those mutations that make a bacteria resistant to a drug, such that we can contemplate therapies that *break* the resistance by annulling the effects of these mutations. If we have a neural network that does any of that well, wouldn't it be great if we could see how it arrives at its conclusions?

Each of the above examples require solutions that fall outside of standard machine learning. Rather than optimally predicting some target value, we are specifically interested learning about the patterns in the data. In this talk I will discuss recent work that allows us to do exactly that. In particular, I will give high-level introductions to DiffNaps (Walter et al. 2024) for learning differential patterns from high-dimensional data, SyFlow (Xu et al. 2024) for learning exceptional subgroups, SySurv for learning survival subgroups. Each is inherently explainable and as they are end-to-end optimizable also highly flexible.

Bio of the presenter
Jilles Vreeken is tenured faculty at the CISPA Helmholtz Center for Information Security, where he leads the Exploratory Data Analysis group. He is also Honorary Professor of Computer Science at Saarland University and an ELLIS Fellow.

His research interests revolve around gaining insights into data and models. This includes developing well-founded theory and efficient methods for learning causal models, informative patterns, as well as robust and explainable machine learning in general. He has authored far too many papers and won too many awards. 

He obtained his Ph.D. in Computer Science in 2009 from Universiteit Utrecht, and then was a post-doctoral researcher at Universiteit Antwerpen until 2013. He joined CISPA in 2018, after having been an independent research group leader at the DFG Cluster of Excellence on Multimodal Computing and Interaction (MMCI) and a Senior Researcher at the Max Planck Institute for Informatics. 

Microkatu area map ,pdf

For further information, please contact Pauli Miettinen, email [email protected].