Plenary Lecture

Title: Mining Big and Complex Data


Increasingly often, data mining has to learn predictive models from big data, which may have many examples or many input/output dimensions and may be streaming at very high rates. Contemporary predictive modeling problems may also be complex in a number of other ways: they may involve (a) structured data, both as input and output of the prediction process, (b) incompletely/partially labelled data, and (c) data placed in a spatio-temporal or network context.

The talk will first give an introduction to the different tasks encountered when learning from big and complex data. It will then
present some methods for solving such tasks, focusing on structured-output prediction, semi-supervised learning (from
incompletely annotated data), and learning from data streams. Finally, some illustrative applications of these methods will be described, ranging from genomics and medicine to image annotation and space exploration

Speaker: Saso Dzeroski


Saso Dzeroski (Sašo Džeroski) is a scientific councillor at the Jozef Stefan Institute and the Centre of Excellence for Integrated Approaches in Chemistry and Biology of Proteins, both in Ljubljana, Slovenia. He is also a full professor at the Jozef Stefan International Postgraduate School and the University of Ljubljana, Faculy of Computer and Information Sciences. His research group investigates machine learning and data mining (including structured output prediction and automated modeling of dynamic systems) and their applications (in environmental sciences, incl. ecology/ecological modelling, and life sciences, incl. systems biology/systems medicine).

The publication record of Professor Džeroski includes 30 volumes (1 co-authored book, 4 co-edited research mnographs, 8 conference proceedings published with reputed publishers, 10 workshop proceedings and 7 journal special issues), more than 40 book chapters, more than 150 journal papers (more than 125 in journals with impact factors), and more than 290 conference papers. The latest two research monographs he has co-edited are »Computational Discovery of Scientific Knowledge« (2007) and »Inductive Databases and Constraint-Based Data Mining« (2010). His work is highly cited, with 14492 citations and h-index = 56 (in Google Scholar on 15 NOV 2017).

He has participated in many international research projects and coordinated three of them in the past. Most recently, he lead the FET XTrack project MAESTRA (Learning from Massive, Incompletely annotated, and Structured Data). He is currently one of the principal investigators in the FET Flagship Human Brain Project. He has been scientific and/or organizational chair of numerous international conferences, including ECML PKDD 2017, DS-2014, MLSB-2009/10, ECEM and EAML-2004, ICML-1999 and ILP-1997/99: ICML and ECML PKDD are two of the most prominent scientific events in the area of machine learning and data science.

Saso Dzeroski received his Ph.D. degree in computer science from the University of Ljubljana in 1995 and was awarded a Jožef Stefan Golden Emblem Prize for his outstanding doctoral dissertation. Immediately thereafter, he received a fellowship from ERCIM, The European Research Consortium for Informatics and Mathematics, awarded to 5% of applicants. He became a fellow of EurAI, the European Association of AI (formerly ECCAI) in 2008, in recognition for his "Pioneering Work in the field of AI and Outstanding Service for the European AI community". In 2015, he was elected a foreign member of the Macedonian Academy of Sciences and
Arts and in 2016 a member of the European Academy (Academia Europea).