June 29, 2020, València, Spain
|Title: Worthy of Trust?|
|16H00||Association Rule Mining with Differential Privacy|
|16H15||Pelican: A Deep Residual Network for Network Intrusion Detection|
Jilles Vreeken, CISPA, Saarland University, GermanyJilles Vreeken is tenured faculty at the Helmholtz Center on Information Security, where he leads the Exploratory Data Analysis group. In addition, he is Honorary Professor at Saarland University and Senior Researcher at the Max Planck Institute for Informatics.
His research interests include data mining, machine learning, and causal inference. He is particularly interested in developing well-founded theory and efficient methods for extracting informative causal models and patterns from large data, and putting these to good use. He has authored 3 book chapters and over 90 conference and journal papers. He received three best paper awards, the ACM SIGKDD 2010 Doctoral Dissertation Runner-Up Award, and the IEEE ICDM 2018 Tao Li Award.
He was a member of the steering committee of ECML PKDD between 2016 and 2019, panel chair for SIAM SDM 2019, tutorial chair for SIAM SDM 2017, program co-chair for ECML PKDD 2016, publicity co-chair for IUI 2015, sponsorship co-chair for ECML PKDD 2014, and workshop co-chair of IEEE ICDM 2012. He co-organised ten workshops and co-lectured six tutorials. He is a member of the editorial board of Data Mining and Knowledge Discovery (DAMI) and of the ECML PKDD Journal Track Guest Editorial Board, in addition he regularly reviews for TKDD, KAIS, JMLR, as well as for KDD, ICDM, NeurIPS, SDM, and ECML PKDD.
He obtained his M.Sc. in Computer Science from Universiteit Utrecht, the Netherlands. He pursued his Ph.D. at the same university under supervision of Arno Siebes, and defended his thesis 'Making Pattern Mining Useful' in 2009. Between 2009 and 2013 he was a post-doctoral researcher at the University of Antwerp, supported by a Post-doctoral Fellowship of the Research Foundation Flanders (FWO). Before joining CISPA in 2018, he was the leader of the Independent Research Group on Exploratory Data Analysis at the DFG Cluster of Excellence on Multimodal Computing and Interaction (MMCI) at Saarland University.
Title: Worthy of Trust?
Abstract: Those in the know have seen it a long time coming, but for everyone else, machine learning suddenly took the world by storm;
algorithms that learn from data can achieve super-human performance, and have inherently change how we do research, make business and policy
decisions, and especially, automate large parts of these. But, is machine learning in general, and deep learning in particular, ready for
prime time yet? Can we trust it to do the right thing, whatever that may be? By Betteridge's law the answer is, of course, no.
Machine learning can fail spectacularly whenever we do not hold it exactly right; in practice machine learning is often very brittle, susceptible to adversarial input, and prone to make highly unfair decisions. Worst of all, methods that provide state-of-the-art performance are inherently unable to articulate why they do what they do, making them very difficult to trust.
In this talk I will discuss the root cause of these problems -- that correlation does not imply causation -- and how we can use causal modelling and robust statistics to achieve machine learning that is indeed worthy of our trust.