UPV University in Spain
Department of Computer Science and Artificial Intelligence
University of the Basque Country UPV/EHU, Spain
Jose A. Lozano received the PhD in computer science in 1998 and currently is full professor in the Department of Computer Science and Artificial Intelligence at the University of the Basque Country UPV/EHU (Spain) where he leads since 2005 the Intelligent Systems Group. His research interests ranged over several areas of computer science, particularly metaheuristic optimization, probabilistic graphical models and machine learning and their application to problems in biomedicine, bioinformatics, ecology and risk analysis, to name a few. Prof. Lozano has published more than 100 ISI journal papers receiving his works more than 9100 citations in google scholar. He is currently associate editor of IEEE Trans. on Neural Networks and Learning Systems and IEEE Trans. Evolutionary Computation and member of the editorial board of Evolutionary Computation journal, memetic computing and several other journals on computational intelligence.
The literature around machine learning has recently seen many problems that depart from the standard supervised classification problem. In these problems the common structure of a supervised dataset where there is a label associated to each instance is broken: an instance can have several labels, a label is assigned to a subset of instances, etc. These problems present different degrees of uncertainty in learning but also in prediction. In this talk we will provide a taxonomy of non-standard machine learning problems illustrating each of them with case examples. We will also elaborate on how to learn classifiers in these scenarios, how to evaluate them and finally we will point out to some results on PAC learning on these problems.
Tilburg School of Economics and Management
Department of Management
Tilburg University, Netherlands
Prof. Michael (Mike) Papazoglou is a highly acclaimed academic with noteworthy experience in areas of education, research and leadership pertaining to computer science, information systems, service-engineering, cloud computing, and digital manufacturing. He holds the Chair of Computer Science and is the executive director of European Research Institute in Service Science (ERISS) at Tilburg University. He is noted as one of the original promulgators of ‘service-oriented computing’ and was the scientific director of the acclaimed EU Network of Excellence in Software Systems and Services (S-CUBE). He is renowned for establishing local ‘pockets of research excellence’ in service science and engineering in several European countries, China, Australia and the Middle East. Papazoglou is an author of the most highly cited papers in the area of service engineering and Web services worldwide with a record of publishing 25 (authored and edited) books, and over 200 prestigious peer-refereed papers along with approx. 17,000 citations (H-index factor 52). He is a distinguished/honorary professor with an exemplary teaching and R&D record at 11 universities around the globe. He is the founder and editor-in-charge of the MIT Press book series on Information Systems as well as the founder and editor-in-charge of the Springer-Verlag book series on Service Science. His expertise is in the areas of Distributed Systems, Service Oriented Computing, Cloud Computing, Data Engineering and Federated Databases, IoT, Software Engineering, Model Driven Architectures, and Smart Applications, such as Smart Cities and Smart Manufacturing.
Smart data systems and applications support the processing and integration of data into a unified view from disparate big data sources, sensors and devices in the Internet of Things, social platforms, and databases, whether on-premises or cloud, and software-as-a- service applications to enable more effective decision making. The decisive criterion here is not necessarily the amount of data available, but smart content techniques that promote not only the collection and accumulation of related data, but also its context, and understanding. This requires discovering associations between the data, prioritizing results, finding useful insights, discovering patterns and trends within the data to reveal a wider picture that is more relevant to the problem in hand and react to them. The mechanisms that convert stale data to smart data focus on knowledge-based meta-data representation techniques that structure and associate the data sets and content, annotate them, link them with associated processes and software services, and deliver or syndicate information to recipients.
Smart Industrial-Purpose Applications are a new generation of software applications that combine the benefits of smart data and advanced analytics to help organizations manage their resources (including humans), data, sensors, processes and systems more efficiently. They promise to bring greater speed and efficiency to industries as diverse as smart agriculture, smart cities, smart manufacturing, and smart healthcare delivery where they can provide meaningful insights to decision makers and help them solve complex problems.
This talk will focus on the role, characteristics, potential of smart data and applications for diverse domains, and their enabling technologies. To illustrate the potential of smart data and applications, the talk will draw on examples that highlight the interplay of medical and technical aspects of smart healthcare applications. Smart healthcare involves deploying computing, information, service, sensor and visualization technologies to aid in preventing disease, improving the quality of care and lowering overall cost. The talk will also examine the design and deployment requirements, particularly for point -of-care medical applications, which emerge from the interplay of the actual clinical situation and the novelty of the smart healthcare application.