Abstract:

Ever more multimedia data gets produced, stored, and shared. This is a well-known phenomenon and quite common for information technology, one might say, but multimedia as a field of computing has always been aiming at humans rather than computers as ultimate consumers: computing was mostly an auxiliary on the path from media creation to human consumption. Despite increasing automation, human consumption is likely to remain the dominating multimedia use case. Since humans have rather fixed sensing and processing capabilities, the dramatic increase in multimedia data production and online availability poses particular “ multimedia big data“ challenges – the more so since the characteristics of multimedia make the well-known „four V“ of big data particularly virulent.

In light of the aforementioned development, the talk will look at big data challenges for multimedia and at upcoming approaches to meeting these challenges. The problem space will be structured according to an imaginary „processing pipeline“ that starts from media capturing via networking and storage/processing until presentation/consumption. Some nonfunctional aspects such as privacy will be addressed, too.

Biography:

Max Mühlhäuser is a Full Professor of Computer Science at Technische Universität Darmstadt, Germany, and head of the Telecooperation Lab. In 1986, he received his Doctorate from the University of Karlsruhe and soon afterwards founded the first European research center for Digital Equipment Corp. (DEC). Since 1989, he worked as either professor or visiting professor at universities in Germany, Austria, France, Canada, and the US. Max published more than 450 articles, co-authored and edited books about Ubiquitous Computing, E-learning, and distributed & multimedia software engineering. Max is deputy speaker of a nationally funded cooperative research center on the Future Internet and directorate member of the Center for Advanced SEcurity research Darmstadt (CASED).

The term „smart grids“ is used to describe the next-generation intelligent energy systems. Smart grids employ state-of-the-art information and communication technology to control generation, distribution and consumption of energy. The degree of information needed on network status is vastly more accurate compared to traditional power networks, and needs to be available in fine granularity in near real-time. The availability of such fine-grained data raises severe privacy concerns in the end-user domain.  For example, the application of non-intrusive load monitoring techniques to high-resolution load profiles allows inferring details on user behavior such as presence, sleep-and-wake cycles and the brands of used appliances. Another challenge in the widespread adoption of smart grid technologies lies in the domain of security. Recent reports of smart meters that can easily be hacked and used to remotely control energy availability in the connected household have not helped to increase user trust. In this talk, the main challenges in the area of smart grid privacy and security from an end-user perspective will be reviewed. At the example of smart metering, selected solutions will be discussed in detail.

Dominik Engel is a professor at the Salzburg University of Applied Sciences in Austria, where he heads the Josef Ressel Center for User-Centric Smart Grid Privacy, Security and Control. He holds a PhD degree in Computer Science from the University of Salzburg. Prior to joining Salzburg University of Applied Sciences, Dominik Engel was a researcher at the Universities of Bremen and Salzburg and product manager at Sony DADC, where he was responsible for video content security. His research interests include smart grid security, multimedia security and technological methods for enhancing end-user trust.

Abstract: Recent trends in large-scale image retrieval consider image representations based on the Bag of Visual Words model, because such representation enables efficient filtering using inverted files, while for more effective retrieval the representation can be enhanced by VLAD or Hemming embedding. In order to improve the effectiveness of the retrieval even more, some models try to relax from Bag of Visual Words model (i.e., from a shared feature space dictionary) and model content of multimedia objects by means of so-called feature signatures. Whereas the effectiveness of models based on feature signatures seems to be promising for several retrieval tasks, the efficiency of the models still represents a serious performance bottleneck, because the models employ expensive adaptive distance measures to compare two feature signatures. In this talk, we present several recent approaches that significantly improve the efficiency of the retrieval when using models based on feature signatures, making the models applicable also for large-scale image retrieval.

CV: Jakub Lokoc received the doctoral degree in software systems from the Charles University in Prague, Czech Republic. He is an assistant professor in the Department of Software Engineering at the Charles University in Prague, Faculty of Mathematics and Physics, Czech Republic. He is a member of siret research group and his research interests include metric access methods, multimedia retrieval and exploration, and similarity modeling.

Der Rückblick zum TEWI-Kolloquium von Martin Varela am 27.10.2014 beinhaltet die Videoaufzeichnung sowie die Folien:

Video

[iframe height=“350″ src=“http://video.aau.at/video.php?video=ftf_valera.mp4″]

Slides

Abstract: In this talk I’ll discuss some ideas about the reasons, both technical and business-related why we care about Quality of Experience (QoE). I’ll present some results related to QoE-driven network and application-level management, and first steps into economically exploiting QoE models, in particular concerning Service Level Agreements, as well as some enablers for doing so.

Bio: Martín Varela received his PhD and MSc from the University of Rennes 1 (Rennes,France), in 2005 and 2002 respectively. He has been an ERCIM fellow, and spent time at SICS and VTT, where he is currently a Principal Scientist. His research interests lie in the QoE domain, with a particular focus on real-time QoE estimates for generic services and applications thereof. He is currently leading VTT’s work on QoE, and is a Finnish management committee member for COST Action IC1003 Qualinet. He has also served as Scientific Coordinator for the Celtic Plus QuEEN project. He is currently co-chair for the IEEE MMTC QoE Interest Group.

Abstract: Streaming data are relevant in finance, computer science, and engineering while they are becoming increasingly important in medicine and biology. In particular, classification, clustering and structural learning of streaming data are receiving increasing attention. These tasks require algorithms and models capable to represent dynamic, sequence and time. Dynamic Bayesian networks and hidden Markov models are used to analyze streaming data. However, these models are concerned with equally spaced time data and thus suffer from several limitations because it is not clear how to discretize timestamps. The talk introduces continuous time Bayesian networks and continuous time Bayesian networks classifiers. Algorithms for parametric and structural learning of continuous time Bayesian network models to solve classification, clustering and structural learning based on multivariate discrete state continuous time trajectories are described. Stationary and non-stationary continuous time Bayesian networks are presented together with their structural learning based on the marginal likelihood approach. Numerical experiments concerning real world applications in finance, biomedicine, neurology and biology are presented.

CV: Prof Fabio Stella is an associate professor at the Dipartimento di Informatica, Sistemistica e Comunicazione of the Università degli Studi di Milano-Bicocca. His research focuses on models and algorithms for data analysis and decision making under uncertainty in the areas of Business Intelligence, Data and Text Mining and Computational Finance. In the winter term 2014/15 he is giving the course 625.605 – Business Intelligence in Klagenfurt.