Michael May
Siemens, Munich, Germany
Artificial Intelligence and the Industrial Knowledge Graph
Abstract
In the context of digitalization Siemens is leveraging various technologies from artificial intelligence and data analytics connecting the virtual and physical world to improve the entire customer value chain. The internet of things has made it possible to collect vast amount of data about the operation of physical assets in real time, as well as storing them in cloud-based data lakes. This rich set of data from heterogeneous sources allows addressing use cases that have been impossible only a few years ago. Using data analytics e.g. for monitoring and predictive maintenance is nowadays in wide-spread use.
We also find an increasing number of use cases based on Deep Learning, especially for imaging applications. In my talk I will argue that these techniques should be complemented by AI-based approaches that have originated in the knowledge representation & reasoning communities.
Especially industrial knowledge graphs play an important role in structuring and connecting all the data necessary to make our digital twins smarter and more effective. The talk gives an overview of existing and planned application scenarios incorporating AI technologies, data analytics and knowledge graphs within Siemens, e.g. building digital companions for product design and configuration or capturing the domain knowledge of engineering experts from service reports using Natural Language Processing.
Short Bio
Michael May is Head of the Technology Field Business Analytics & Monitoring at Siemens Corporate Technology, Munich, and responsible for eleven research groups in Europe, US, and Asia. He is driving research at Siemens in data analytics, machine learning and big data architectures. In the last two years he was responsible for creating the Sinalytics platform for Big Data applications across Siemens’ business. Before joining Siemens in 2013, Michael was Head of the Knowledge Discovery Department at the Fraunhofer Institute for Intelligent Analysis and Information Systems in Bonn, Germany. In cooperation with industry he developed Big Data Analytics applications in sectors ranging from telecommunication, automotive, and retail to finance and advertising.
Between 2002 and 2009 Michael coordinated two Europe-wide Data Mining Research Networks (KDNet, KDubiq). He was local chair of ICML 2005, ILP 2005 and program chair of the ECML/PKDD Industrial Track 2015. Michael did his PhD on machine
discovery of causal relationships at the Graduate Programme for Cognitive Science at the University of Hamburg.
Jean-Marc Petit
INSA Lyon and Université de Lyon, France
Bridging the Gap between Data Diversity and Data Dependencies
Abstract
Data dependencies are declarative statements allowing to express constraints. They turn out to be useful in many applications, for example from database design (functional, inclusion, multi-valued, … dependencies) to data quality (conditional functional dependencies, matching dependencies, denial dependencies, …). Their practical impacts in many commercial tools acknowledge their importance and utility. Specific data dependencies have been proposed to take into account data diversity encountered in practice, i.e. inconsistency, uncertainty, heterogeneity…
In this talk, I will introduce the main ingredients required to unify most of data dependencies proposed in the literature. Two approaches will be presented: The first one is a declarative query language, called RQL, which is a user-friendly SQL-like query language devoted to data dependencies. The second one is to study structural properties on data domains to define data dependencies through a lattice point of view.
Short Bio
Jean-Marc Petit received his PhD in Computer Sciences in 1996 from Université Lyon 1. From 1997 to 2005, he was an Associate Professor at the Université Blaise Pascal, Clermont-Ferrand, France. In 2005, he was appointed as a full Professor at INSA Lyon and LIRIS laboratory (UMR 5205 CNRS). From 2008 to 2015, he led the database group (more than 30 researchers and PhD students) at LIRIS. From 2007 to 2015, he led (as director) the master by research program in Computer Sciences. He was involved in the organization of many conferences, including VLDB 2009 (held in Lyon, France). He was president of the French learned society of informatics (Société Informatique de France – SIF) from 2015 to 2018. Since 2015, he is the deputy chair of the LIRIS Laboratory. He is involved in many program committees of international conferences and in many academic and industrial projects. His main research interests deal with databases and data mining.
Sašo Džeroski
Jozef Stefan Institute and
Jozef Stefan International Postgraduate School
Slovenia
Mining Big and Complex Data
Abstract
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 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.
Short Bio
Sašo Džeroski is a scientific councillor at the Jozef Stefan Institute and a full professor at the Jozef Stefan International Postgraduate School. He leads a research group which develops methods for machine learning and data mining (including structured output prediction and automated modeling of dynamic systems) and investigates their use (in environmental sciences, incl. ecology/ecological modelling, and life sciences, incl. systems biology/systems medicine). Hispublication record 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 300 conference papers.
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 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. He became a fellow of EurAI, the European Association of Artificial Intelligence (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 Academia Europea (European Academy of Humanities, Letters and Sciences).