Feature Selection and Knowledge Discovery in AutoML and Automated Causal Discovery
Ioannis Tsamardinos
Date: Tuesday, Sep 12, 2023, 14:30 - 15:00
Abstract:
It is often the case that the primary goal of an analysis is not model construction, but discovering new knowledge in the form of the features that determine the predictions. In this talk, we’ll get an in-depth analysis of the feature selection problem, the multiple feature selection problem, and feature importance and how they can be solved with scalable algorithms in the context of AutoML. We discuss the connection of the feature selection problem with the causal generative mechanism of the data and describe our efforts to build an Automated Causal Discovery engine. Finally, we’ll present use-cases and results from our AutoML system called JADBio.
Bio:
Ioannis Tsamardinos, Ph.D., is a Professor at the Computer Science Department of the University of Crete, CSO, and co-founder of JADBio (Gnosis Data Analysis PC), a University start-up. He obtained his Ph.D. from the Intelligent Systems Program at the University of Pittsburgh in 2001. He then worked as an Assistant Professor at the Department of Biomedical Informatics at Vanderbilt University until 2006 when he returned to Greece. Prof. Tsamardinos’ main research directions include machine learning, bioinformatics, and artificial intelligence. More specifically his work emphasizes automated machine learning, feature selection, and causal discovery. Prof. Tsamardinos has over 140 publications in international journals, conferences, and books. Distinctions with colleagues and students a Gold Medal in the Student Paper Competition in MEDINFO 2004, the Outstanding Student Paper Award in AIPS 2000, the NASA Group Achievement Award for participation in the Remote Agent team, and others. Statistics on recognition of work include more than 11000 citations (1100+ a year), and h-index of 43 (as estimated by Google Scholar). Ioannis has been awarded the European and Greek national grants of excellence, the ERC Consolidator, ERC Proof of Concept, and the ARISTEIA II grants respectively.