Representation and Learning of Structured Dynamic Bayesian Models in Non Stationary Environments
Carlo Regazzoni, University of Genova, Italy
Directions in Security Research
Jan Camenisch, , Switzerland
Speeding up the execution of numerical computations and simulations with rCUDA
Jose Duato, Independent Researcher, Spain
Advances and Future Challenges in Machine Learning and Knowledge Extraction
Andreas Holzinger, Medical University Graz, Austria
Representation and Learning of Structured Dynamic Bayesian Models in Non Stationary Environments
Carlo Regazzoni
University of Genova
Italy
![]()
Brief Bio
Carlo S. Regazzoni (Ph.D.) is full professor of Cognitive Telecommunications Systems at DITEN, University of Genova, Italy. His main research interests include (see www.isip40.it) cognitive dynamic systems, adaptive and self-aware video processing, tracking and recognition, generative models and inference schemes based on hierarchical dynamic Bayesian networks, software and cognitive radio. He has been responsible of several national and EU funded research projects. He is currently coordinator of international PhD courses on Interactive and Cognitive Environments involving several European universities. He is author of peer-reviewed papers on international journals (90) and international conferences/books (350). He served as general chair (IEEE AVSS2009), technical program chair (IEEE ICIP2005, NSIP2002), associate editor (IEEE Trans on Image Processing, IEEE Trans on Mobile Computing, et al.), guest editor (Proceedings of the IEEE, IEEE Signal Processing Magazine et al.) in international conferences and Journals. He has served in many roles in governance bodies of IEEE SPS. He is currently serving as Vice President Conferences IEEE Signal Processing Society in 2015-2017.
Abstract
A large part of current research in ICT is centred on terms like Internet of Things or Big Data, that are focused on the benefits that interconnected smart machines and intelligent industrial processes can have in our society. Enabling techniques that are underlying such processes aim towards an increased capability of adaptive and autonomous automation of physical and logical “Things”, that has as implication the definition of new computational paradigms and frameworks. A key role is here represented by probabilistic signal processing architectures and techniques for representation and learning. Such techniques have a sufficient maturity to be good candidates for capturing generality, self-awareness, adaptability, flexibility and reconfigurability through experience based learning that is needed to make it possible “Things” to adapt their behaviors in non stationary environments where they will have to operate. In this talk, Dynamic Bayesian models and related machine learning techniques will be presented and discussed. It will be shown that such techniques could be used within a Cognitive Dynamic System framework associated with “Ego-Things” i.e. self-aware smart objects adaptively performing computation driven tasks associated with their physical or logical actuation capability in non stationary environments. Applications examples will be presented dealing with crowd surveillance in smart buildings, and, more in general, with dynamic functionalities in cognitive environments composed by smart ego-objects.
Directions in Security Research
Jan Camenisch
Switzerland
Brief Bio
Dr. Jan Camenisch is a Principal Research Staff Member at IBM Research - Zurich and leads the Privacy & Cryptography research team. He's a member of the IBM Academy of Technology and an IEEE Fellow. He is a leading scientist in the area of privacy and cryptography, has published over 100 widely cited papers, and has received a number of awards for his work, including the 2010 ACM SIGSAC outstanding innovation award and the 2013 IEEE computer society technical achievement award.
Jan is also a co-inventor of Identity Mixer, a unique cryptographic protocol suite for privacy-preserving authentication and transfer of certified attributes.
Jan previously led the FP7 European research consortia PRIME and PrimeLife, and he and his team continue to participate in many other projects including ABC4Trust, AU2EU, and Witdom. Jan currently holds an Advanced ERC grant for "Personal Cryptography".
Abstract
Our digital environment and the way we use it change rapidly. This poses a number of new security and privacy challenges and amplifies many known issues. In this talk, we identify and discuss some of these challenges. We will then assess how and to what extend it would be possible to address these challenges today, identify some gaps and then provide future research directions towards closing these gaps.
Speeding up the execution of numerical computations and simulations with rCUDA
Jose Duato
Independent Researcher
Spain
![]()
Brief Bio
Jose Duato is Professor in the Department of Computer Engineering (DISCA) at the Technical University of Valencia.His current research interests include interconnection networks, on-chip networks, and multicore and multiprocessor architectures. He published over 500 refereed papers. According to Google Scholar, his publications received more than 12,000 citations. He proposed a theory of deadlock-free adaptive routing that has been used in the design of the routing algorithms for the Cray T3E supercomputer, the on-chip router of the Alpha 21364 microprocessor, and the IBM BlueGene/L supercomputer. He also developed RECN, a scalable congestion management technique, and a very efficient routing algorithm for fat trees that has been incorporated into Sun Microsystem's 3456-port InfiniBand Magnum switch. Prof. Duato led the Advanced Technology Group in the HyperTransport Consortium, and was the main contributor to the High Node Count HyperTransport Specification 1.0. He also led the development of rCUDA, which enables remote virtualized access to GP-GPU accelerators using a CUDA interface.Prof. Duato is the first author of the book "Interconnection Networks: An Engineering Approach". He also served as a member of the editorial boards of IEEE Transactions on Parallel and Distributed Systems, IEEE Transactions on Computers, and IEEE Computer Architecture Letters. Prof. Duato was awarded with the National Research Prize in 2009 and the “Rey Jaime I” Prize in 2006.
Abstract
This keynote will present some techniques to speed up the execution of numerical computations and simulations, with special emphasis on the use of hardware accelerators. The talk will present the architecture of the most popular accelerators currently in use, the implications for the programmers, and the main limitations of the current commercial implementations. The talk will also describe a recent technology for virtualizing accelerators that dramatically improves the utilization and effective computing power of those accelerators while reducing power consumption. Finally, the measured benefits in some real computing installations will be shown.
Advances and Future Challenges in Machine Learning and Knowledge Extraction
Andreas Holzinger
Medical University Graz
Austria
Brief Bio
Andreas Holzinger is lead of the Holzinger Group HCI–KDD, Institute for Medical Informatics & Statistics at the Medical University Graz, and Associate Professor of Applied Computer Science at the Faculty of Computer Science and Biomedical Engineering at Graz University of Technology. Currently, Andreas is Visiting Professor for Machine Learning in Health Informatics at the Faculty of Informatics at Vienna University of Technology. He serves as consultant for the Canadian, US, UK, Swiss, French, Italian and Dutch governments, for the German Excellence Initiative, and as national expert in the European Commission. Andreas obtained a PhD in Cognitive Science from Graz University in 1998 and his Habilitation (second PhD) in Computer Science from Graz University of Technology in 2003. Andreas was Visiting Professor in Berlin, Innsbruck, London (twice), and Aachen. Andreas and his Group work on extracting knowledge from data and foster a synergistic combination of methodologies of two areas that offer ideal conditions towards unraveling problems with complex health data: Human-Computer Interaction (HCI) and Knowledge Discovery/Data Mining (KDD), with the central goal of supporting human intelligence with machine learning to discover novel, previously unknown insights into data. To stimulate crazy ideas at international level without boundaries, Andreas founded the international Expert Network HCI–KDD. Andreas is Associate Editor of Knowledge and Information Systems (KAIS), Associate Editor of Springer Brain Informatics (BRIN) and Section Editor of BMC Medical Informatics and Decision Making (MIDM). He is member of IFIP WG 12.9 Computational Intelligence, the ACM, IEEE, GI and the Austrian Computer Society. Home: http://hci-kdd.org
Publications see <link> https://scholar.google.com/citations?hl=en&user=BTBd5V4AAAAJ&view_op=list_works&sortby=pubdate
Abstract
Today the problem are heterogeneous, probabilistic, high-dimensional and complex data sets. The challenge is to learn from such data to extract and discover knowledge, and to help to make decisions under uncertainty. In automatic machine learning (aML) great advances have been made, for example, in speech recognition, recommender systems, or autonomous vehicles. Automatic approaches greatly benefit from "big data" with many training sets. However, sometimes we are confronted with a small amount and complex data sets, where aML suffers of insufficient training samples. The application of such aML approaches in complex application domains such as health informatics seems elusive in the near future, and a good example are Gaussian processes, where aML (e.g. standard kernel machines) struggle on function extrapolation problems which are trivial for human learners. In such situations, interactive Machine Learning (iML) can be beneficial where a human-in-the-loop helps in solving computationally hard problems, e.g., subspace clustering, protein folding, or k-anonymization of health data, where the knowledge and experience of human experts can help to reduce an exponential search space through heuristic selection of samples. Therefore, what would otherwise be an NP-hard problem reduces greatly in complexity through the input and the assistance of a human agent involved in the learning phase. ML is a fast growing and very practical field with many business applications and much open research challenges, particularly in multi-task learning, transfer learning and hybrid multi-agent systems with humans-in-the-loop. Consequently, successful ML needs a concerted effort, fostering integrative research between experts ranging from diverse disciplines from data science to visualization and tackling complex challenges needs both disciplinary excellence and a cross-disciplinary skill-set and international joint work without any boundaries.