Fundamental Courses - Summer Semester
Visiting Professor Courses - Summer Semester
195.092 Algorithm Design in Different Models of Computation (Prof. Guy Even, Tel-Aviv University)
195.072 Current Trends in Computer Science (Ring lecture with talks given by the visiting professors of the current summer term)
From Web Data to Network Analysis (course details to be confirmed)
Fundamental Courses - Winter Semester
Fundamental Courses - Winter Semester
Visiting Professor Courses - Winter Semester
195.084 Graph Partitioning and Graph Clustering in Theory and Practice (Dr. Christian Schulz, Karlsruhe Institute of Technology, Germany)
195.085 Domain and Requirements: Science & Engineering (Prof. Dines Bjørner, Technical University of Denmark)
Fundamental Courses - Summer Semester
Visiting Professor Courses - Summer Semester
195.082 Automata-theoretic methods for infinite-state verification (Prof. Tomáš Vojnar, Brno University of Technology, Czech Republic)
Fundamental Courses - Winter Semester
Invited talks within the lecture series on current trends in computer science
The lecture series "Current Trends in Computer Science" will be offered for all computer science students. In this lecture series, the visiting professors of the PhD School will give a presentation about their research area.
30.03.2017: At the Core of Computing: Operational (Re-)Construction
Speaker: Christiane Floyd
Date: 30.03.2017, 14:00 c.t.
Location: EI5 Hochenegg, Gusshausstrasse 25-29
The notion of 'Computing' covers scientific fields like ‘informatics’, ‘computer science’, as well as the practice of developing computing systems and artifacts in both research and industry. ‘To be computed’ does not refer to calculation only, but to reframing a problem so as to make it accessible to operationalization. Computing systems and artifacts are so pervasive in society, they exhibit such an enormous diversity and transform our lives in so widely divergent fields that the question arises: what ties together the scientific disciplines underlying their construction and how do they relate to other sciences?
In this talk, we will be concerned with two questions:
1. What is the place of Computing in the landscape of the sciences?
2. How can we characterize Computing in the field of tension between the formal, the technical and the social world?
It will be argued that Computing comes with a thinking style – operational (re-)construction – which can be applied to most any area of interest and will be explored in some depth in the talk.
Christiane Floyd is Professor emerita for Software Engineering of the University of Hamburg and Honorary Professor at TU Wien. She was the first female professor for Computer Science in Germany. Floyd did her Ph.D. in Mathematics at the University of Vienna in 1966. From 1966 to 1968 she worked as a software engineer for Siemens, Munich. From 1968 to 1973 she was working and teaching at Stanford University. She was also senior consultant for Softlab, Munich (1973-1977). Floyd served as professor at TU Berlin (1978-1991) and University of Hamburg (1991- 2008). Together with her research group STEPS she developed a participative and evolutionary approach for software development, which laid the epistemological foundation for the field of software engineering. In 1987 she joined the board of FIfF (Forum InformatikerInnen für Frieden und gesellschaftliche Verantwortung) as a founding member and chair. In this position Floyd worked on the frontier for increasing the societal responsibility of Computer Science. In 2000 she acted as dean for the research area ICT within the “Internationale Frauenuniversität“. Since 2006 she has furthermore been active in the field of ICT for Development in Ethiopia.
30.06.2016 Model checking and system design as a machine learning problem
Constructing an efficient statistical learning approach
Speaker: Guido Sanguinetti, University of Edinburgh
Date: 30.06.2016, 14:00
Location: Informatik Hörsaal
I will consider the problem of probabilistic model checking of temporal logic formulae on stochastic processes with parametric uncertainty. This is computationally challenging, as in principle one needs an expensive model checking step for each value of the model parameters. We show that the dependence of truth probabilities on parameters is differentiable, which allows us to construct an efficient statistical learning approach to solve the problem: we collect a sparse sample of parameter/ truth probability pairs, and treat them as a training set for a machine learning predictor based on Gaussian Processes. Due to the smoothness result, this approach converges to the true value in the large sample size. I will further discuss how these ideas can be generalised in a system design/ identification context.
Guido Sanguinetti is a Reader in Informatics at the University of Edinburgh, UK. His interests focus on machine learning methodologies for complex systems, in particular dynamical biological systems. He has published >60 research articles in leading international journals including Science, PNAS, Nature Communications. He is in receipt of an ERC Starting Grant and was awarded the 2012 PNAS Cozzarelli Prize in Engineering and Applied Science.
07.06.2016 Model Theory in Computer Science – Recurrent themes and some lessons learned
Speaker: Johann A.Makowsky, Technion – Israel Institute of Technology
Date: 07.06.2016, 14:00
I give an updated version of my talk given at CSL-2011 (retiring EACSL-president's address) with the same title. I describe my research from hard core model theory to soft model theory and my transition to logic for computer science and then to logic applied to combinatorics in a leasurely way. I also add some lessons I learned about research, human memory, and science as a cultural system. It is planned for a wide audience.
Johann A. Makowsky is a distinguished mathematician working in mathematical logic and the logical foundations of computer science and combinatorics. He studied at the Swiss Federal Institute of Technologyfrom 1967-73 and held visiting positions at the Banach Center in Warsaw, Stanford University, Simon Fraser University (Canada), University of Florence, MIT, Lausanne University, and ETH Zurich. Prof. Makowsky is full professor at the Technion in Haifa. He also was a founding member of the European Association of Computer Science Logic in 1992, its vice-president (2002-2004) and president (2004-2009), and was a member of EACSL's executive council until 2014. His research includes contributions in model theory, database theory, and logic programming.
06.06.2016, Vote Counting as Mathematical Proof
Speaker: Dirk Pattinson, Australian National University
Date: 06.06.2016, 14:00
We give an overview of the current state of electronic voting as used for legally binding elections throughout the world. Of particular interest is the question of trust: what mechanisms are in place (or should be) so that the various stakeholders are in fact convinced that the final outcome of an electronic election is in fact consistent with the procedures outlined in law? We then specialise this question to the question of electronic vote counting, and present an approach to universal verifiability that enables a wide range of members of the general public to convince themselves of the correctness of ballot counting.
After receiving MSc and PhD from Ludwig-Maximilians Universität, München, Dirk held academic positions at the University of Leicester and Imperial College London (both UK) before taking up his present position of Associate Professor at the Australian National University in Canberra, Australia.
01.06.2016 Do you want a Robot Lover? The Ethics of Caring Technologies
Speaker: Blay Whitby, University of Sussex
Date: 01.06.2016, 14:00
Do You Want a Robot Lover? You might perhaps think that you do and that it is nobody else's business but yours, but the widespread use of robots in intimate and caring roles will bring about important social changes. We need to examine these changes now and consider them from an ethical standpoint. Robotic carers and artificial companions are a technology that is likely to be available in the near to mid‐term future. In Japan and South Korea robots are seen as potential carers for the elderly and as babysitters. Many researchers are looking to make their products display emotion and respond to emotional displays by users. At least one writer has predicted marriage to robots will be accepted in progressive countries by 2050. This talk will examine some of the implications of these ‐ both technical and social. Do they represent socially and ethically acceptable developments? What is likely to be technically feasible and just what should we allow?
Dr. Blay Whitby is a philosopher and ethicist concerned with the social impact of new and emerging technologies. He is a leading researcher in the field and the author of many books, chapters and papers on the subject including “On Computable Morality”, “Reflections on Artificial Intelligence: The Legal, Moral and Ethical Dimensions" and “Artificial Intelligence, A Beginner’s Guide”.
18.05.2016 On the Evaluation of Unsupervised Outlier Detection - Measures, Datasets, and an Empirical Study
Speaker: Arthur Zimek, Ludwig-Maximilians-Universität München
Date: 18.05.2016, 16:00
Location: GM2 Radinger Hörsaal
The evaluation of unsupervised outlier detection algorithms is a constant challenge in data mining research. Little is known regarding the strengths and weaknesses of different standard outlier detection models, and the impact of parameter choices for these algorithms. The scarcity of appropriate benchmark datasets with ground truth annotation is a significant impediment to the evaluation of outlier methods. Even when labeled datasets are available, their suitability for the outlier detection task is typically unknown. Furthermore, the biases of commonly-used evaluation measures are not fully understood. It is thus difficult to ascertain the extent to which newly-proposed outlier detection methods improve over established methods. In this talk, we discuss our recent research results: we performed an extensive experimental study on the performance of a representative set of standard k nearest neighborhood-based methods for unsupervised outlier detection, across a wide variety of datasets prepared for this purpose. Based on the overall performance of the outlier detection methods, we provide a characterization of the datasets themselves, and discuss their suitability as outlier detection benchmark sets. We also examine the most commonly-used measures for comparing the performance of different methods, and suggest adaptations that are more suitable for the evaluation of outlier detection results.
Dr. Arthur Zimek is a Privatdozent in the database systems and data mining group at the Ludwig-Maximilians-Universität München (LMU), Germany. 2012-2013 he was a postdoctoral fellow in the department for Computing Science at the University of Alberta, Edmonton, Canada. In 2014 he was a visiting professor at Technical University Vienna, Austria. He holds degrees in bioinformatics, philosophy, and theology, involving studies at universities in Munich, Mainz (Germany), and Innsbruck (Austria) and finished his Ph.D. thesis in informatics on "Correlation Clustering" at LMU in summer 2008. For this work, Zimek received the "SIGKDD Doctoral Dissertation Award (runner-up)" in 2009. His research interests include ensemble techniques, clustering, and outlier detection, methods as well as evaluation, and high dimensional data. Zimek published more than 60 papers at peer reviewed conferences and in international journals. Together with his co-authors, he received the "Best Paper Honorable Mention Award" at SDM 2008 and the "Best Demonstration Paper Award" at SSTD 2011. Zimek has been a member or senior member of program committees of the leading data mining conferences (e.g. SIGKDD, ECMLPKDD, CIKM, SDM) and serves as reviewer for journals like ACM TKDD, IEEE TKDE, Data Mining and Knowledge Discovery (Springer), Machine Learning (Springer).
03.05.2016 Search-Based Software Engineering, Foundations, Challenges and Recent Advances
Speaker: Marouane Kessentini, University of Michigan
Date: 03.05.2016, 15:00
A growing trend has begun in recent years to move software engineering problems from human-based search to machine-based search that balances a number of constraints to achieve optimal or near-optimal solutions. As a result, human effort is moving up the abstraction chain to focus on guiding the automated search, rather than performing the search itself. This emerging software engineering paradigm is known as Search Based Software Engineering (SBSE). It uses search based optimization techniques, mainly those from the evolutionary computation literature to automate the search for optimal or near-optimal solutions to software engineering problems. The SBSE approach can and has been applied to many problems in software engineering that span the spectrum of activities from requirements to maintenance and reengineering. Several challenges have to be addressed to mainly tackle the growing complexity of software systems nowadays in terms of number of objectives, constraints and inputs/outputs. Most of software engineering problems are multi- and many-objective by nature to find a trade-off between several competing goals. In addition, several software engineering solutions lack robustness due to the dynamic environments of software systems (e.g., requirements change over time). Furthermore, it is essential to understand the points at which human oversight, intervention, resumption-of-control and decision making should impinge on automation. In this talk, I will give, first, an overview about SBSE then I will focus on some contributions that I proposed, along with my research group and my industrial partners. Finally, I will discuss possible new research directions in SBSE.
Dr. Marouane Kessentini is an Assistant Professor in the Department of Computer and Information Science at the University of Michigan. He is the founder of the Search-Based Software Engineering (SBSE) research lab. He has several collaborations with different industrial companies on the use of computational search, machine learning and evolutionary algorithms to address several software engineering problems such as software quality, software testing, software migration, software evolution, etc. He received a best PhD award from University of Montreal in 2012 and a Presidential BSc Award from the President of Tunisia in 2007. He received many grants from both industry and federal agencies and published around 70 papers in software engineering journals and conferences, including 3 best paper awards. He is also the founder of the North American Symposium on Search Based Software Engineering, funded by the National Science Foundation (NSF), and a guest editor of the first Special Issue on Search Based Software Engineering at the IEEE Transactions on Evolutionary Computation Journal (2016). He is an invited speaker in the 2016 IEEE World Congress on Computational Intelligence (Vancouver, Canada).
28.01.2015 Big Data Graph Algorithms
Speaker: Christian Schulz, Karlsruhe Instititute of Technology
Date: 28.01.2015, 13:00
The talk covers recent work on big data graph algorithms. We touch three different areas: graph partitioning, graph drawing and kernelization techniques for the independent set problem in practice. In particular, we present parallel partitioning algorithms that are able to process huge complex networks which have recently attracted considerable interest. We engineer parallel size-constraint graph clustering algorithms that are used as a subroutine at multiple stages of multilevel graph partitioning. The resulting system is more scalable and achieves higher quality than other state-of-the-art systems. As an example, we partition a web graph with 3.3G edges in 16 seconds on a high-performance cluster and compute a high quality partition -- none of the competing systems can handle this graph. In another line of research, we compute graph layouts using a multilevel algorithm with respect to a recently proposed metric that combines layout stress and entropy. Drawing large graphs appropriately is an important step for the visual analysis of data from real-world networks. The proposed system uses a simple local iterative scheme within a multilevel approach, approximates long-range forces and uses shared-memory parallelism. In comparison to the previously best maxent-stress optimizer, which is sequential, our parallel implementation is on average 30 times faster while producing comparable solution quality. Lastly, we develop an advanced evolutionary algorithm, which incorporates kernelization techniques to compute large independent sets in huge sparse networks. The key idea is to apply an evolutionary algorithm on a problem kernel, and then proceed recursively with inexact kernelization techniques. This opens up the reduction space, which not only speed up the computation of large independent sets drastically, but also enabled us to compute high-quality independent sets on much larger instances than previously reported in the literature.
Christian Schulz is currently a visiting researcher at the TU Wien and a postdoctoral researcher at the Karlsruhe Institute of Technology (KIT). He obtained his master degrees in mathematics and computer science as well as his Ph.D with summa cum laude at the KIT. He was the best student of the year 2010 of the faculty of informatics of the KIT and received multiple awards for his dissertation. His graph partitioning algorithms -- KaHIP -- have been able to improve or reproduce most of the benchmark entries in the Walshaw Benchmark and scored most of the points in the 10'th DIMACS Implementation Challenge on Graph Partitioning and Clustering.
29.10.2015 A New Foundation for Computing Science
Speaker: Dines Bjørner, Technical University of Denmark
Date: 29.10.2015, 14:00
Location: HS1, Theresianumgasse 27
We argue that computing systems requirements must be based on precisely described domain models. We further argue that domain science and engineering offers a new dimension in computing. We review our work in this area and we hint at a research and experimental engineering programme for the first two phases of the triptych of domain engineering, requirements engineering, and software design.
Dines Bjørner is professor emeritus at the Technical University of Denmark in Lyngby. He was also responsible for establishing the United Nations University International Institute for Software Technology (UNU-IIST) in Macau. His research deals with domain engineering, requirements engineering, and formal methods. Moreover, he worked on the Vienna Development Method (VDM) at IBM in Vienna as well as being involved in realizing the RAISE (Rigorous Approach to Industrial Software Engineering) formal method with tool support.
10.06.2015 From Pointers to List Containers
Speaker: Tomáš Vojnar, Brno University of Technology, Czech Republic.
Attention: Talk had to be postponed!
Date: 10.6.2015, 16:00
We propose a method that transforms a C program manipulating lists via low-level pointer statements into an equivalent program where the lists are manipulated via calls of standard high-level container operations like push_back, pop_front, etc. The input of our method is a C program annotated by a special form of shape invariants which can be obtained from current automatic shape analysers. The resulting program where low-level pointer statements are summarized into high-level container operations is more understandable and better suitable for program analyses since the burden of analysing low-level pointer manipulations gets removed. We have implemented our approach and successfully tested it through a number of experiments, including experiments showing that our method can indeed be used to ease program analysis by separating shape analysis from analysing data-related properties. The result is a joint work with Kamil Dudka, Lukas Holik, and Petr Peringer from Brno University of Technology and Marek Trtik from Verimag, Grenoble.
Tomas Vojnar received his PhD at the Brno University of Technology (BUT) in 2001. From 2001 to 2003, he worked as a postdoc researcher at LIAFA, Université Paris Diderot. Since 2003, he works at the Faculty of Information Technology of BUT. He defended his habilitation thesis in 2007 and became a full professor in 2012. His research focuses mainly on automated verification, especially over infinite-state and concurrent systems. In the former area, his work has included formal verification of parameterized systems, abstract regular model checking, and shape analysis of programs with pointers and dynamic linked data structures. He has also worked on efficient ways of dealing with various kinds of non-deterministic automata and logics, with applications in the above mentioned verification approaches. As for concurrent programs, he has mainly worked in the area of their noise-based testing and dynamic analysis, self-healing, as well as applications of search techniques in noise-based testing. He has (co-)authored over 100 scientific publications as well as multiple software tool prototypes (e.g., Forester, Predator, SearchBestie, AnaConDA, Spen, or Slide). These works won multiple best paper awards, such as the EATCS Best Theory Paper Award of ETAPS'10, the best tool paper of RV'12, as well as multiple medals in the international software verification competition SV-COMP'12-15, including a Goedel medal at the FLoC Olympic Games 2014.
09.06.2015 Leveraging Computational Social Science to address Grand Societal Challenges
Speaker: Noshir S. Contractor, Northwestern University, USA.
Date: 09.06.2015, 17:00
The increased access to big data about social phenomena in general, and network data in particular, has been a windfall for social scientists. But these exciting opportunities must be accompanied with careful reflection on how big data can motivate new theories and methods. Using examples of his research in the area of networks, Contractor will argue that Computational Social Science serves as the foundation to unleash the intellectual insights locked in big data. More importantly, he will illustrate how these insights offer social scientists in general, and social network scholars in particular, an unprecedented opportunity to engage more actively in monitoring, anticipating and designing interventions to address grand societal challenges.
Noshir Contractor is the Jane S. & William J. White Professor of Behavioral Sciences in the McCormick School of Engineering & Applied Science, the School of Communication and the Kellogg School of Management at Northwestern University, USA. He is the Director of the Science of Networks in Communities (SONIC) Research Group at Northwestern University and a board member of the Web Science Trust. He is investigating factors that lead to the formation, maintenance, and dissolution of dynamically linked social and knowledge networks in a wide variety of contexts.
His research program has been funded continuously for almost two decades by major grants from the U.S. National Science Foundation with additional funding from the U.S. National Institutes of Health (NIH), Army Research Laboratory, Air Force Research Laboratory, Army Research Institute, NASA, Rockefeller Foundation, Gates Foundation, and the MacArthur Foundation.
His book titled Theories of Communication Networks (co-authored with Professor Peter Monge and published by Oxford University Press), received the 2003 Book of the Year award from the Organizational Communication Division of the National Communication Association. He was a recipient of the 2014 National Communication Association Distinguished Scholar Award and in 2015 he was elected a Fellow of the International Communication Association. He is also the co-founder and Chairman of Syndio, which offers organizations products and services based on network analytics.
Professor Contractor has a Bachelor’s degree in Electrical Engineering from the Indian Institute of Technology, Madras and a Ph.D. from the Annenberg School of Communication at the University of Southern California.
01.06.2015 Image processing applications in small and big scales
Speaker: László G. Nyúl, Szeged University, Hungary.
Date: 01.6.2015, 14:00
The main task of digital image processing is to process and analyze image-like data. Data from any source can be understood as an image if it can be represented in a matrix form. Most often we deal with images of 2 or 3 spatial dimensions, but time (e.g. in video sequences) and spectral dimensions (e.g. in aerial remote sensing) can also be handled. Among others, image processing deals with image restoration (the compensation for degradation), image segmentation (the delineation of objects depicted in the image), image registration (determining the spatial correspondence between images acquired at different time and/or space), image fusion (forming a new image by combining the information from multiple images), calculating object properties and describing the objects and/or the scene depicted in the image. Image processing is applied in an increasing number of areas in life, such as in medical image diagnosis systems, industrial manufacturing and quality control, security systems, from autonomous robots to smartphones with cameras. In this lecture the audience shall get a taste of these topics and challenges through selected projects from the half century history of image processing at the University of Szeged.
László G. Nyúl has received his degrees in mathematics and computer science from the University of Szeged, Hungary. Currently he is the Head of the Department of Image Processing and Computer Graphics, Faculty of Science and Informatics, University of Szeged. Between 2011 and 2014 he has been deputy head of the Institute of Informatics at the University of Szeged. His research interest is digital image processing with a focus on medical applications and fuzzy approaches. Since 1991 he lectures at the University of Szeged in various subjects, including Computer graphics, Graphical systems, Advanced graphics algorithms, Multimedia, Image processing, Fuzzy techniques in image processing, Image databases, and Medical imaging.
12.05.15 Search for the Science of Computing: Competing Viewpoints
Speaker: Matti Tedre, University of Stockholm, Sweden.
Date: 12.5.2015, 16:00
Location: HS 8
Computing as a discipline started form up after a new era of computing dawned in the 1930s and 1940s with a number of technical and theoretical innovations. Clashes between different agenda for the nascent field characterized computing's disciplinary debates from the beginning. One of the early dilemmas was to describe computing as a rigorous, mathematically oriented field separate from mathematics. The software crisis brought up arguments for the importance of software engineering as well as studying people, the users of computer systems. All along computing was also characterized as an empirical science - first by virtue of its theoretical foundations, and then by virtue of its aims, methods, subject matter, and computational modeling's predictive success in other fields. This talk discusses some central debates about computing's disciplinary identity and their importance for the discipline's formation.
Matti Tedre is the author of The Science of Computing: Shaping a Discipline (Taylor & Francis / CRC Press, 2014). He's an associate professor at Stockholm University, and an adjunct professor in several other universities. His main fields of interest are the history and philosophy of computer science and computer science education research.
13.04.15 Real-time Model Predictive Control for the Optimal Charging of a Lithium-ion Battery
Speaker: Davide M. Raimondo, University of Pavia, Italy.
Date: 13.4.2015, 14:00
Location: EI 9
Li-ion batteries are widely used in industrial applications due to their high energy density, slow material degradation, and low self-discharge. The existing advanced battery management systems (ABMs) in industry employ semi-empirical battery models that do not use first-principles understanding to relate battery operation to the relevant physical constraints, which results in conservative battery charging protocols. In this talk we present a Quadratic Dynamic Matrix Control (QDMC) approach to minimize the charge time of batteries to reach a desired state of charge (SOC) while taking temperature and voltage constraints into account. This algorithm is based on an input-output model constructed from a first-principles electrochemical battery model known in the literature as the pseudo two-dimensional (P2D) model. In simulations, this approach is shown to significantly reduce charging time.
Davide M. Raimondo was born in Pavia, Italy, in 1981. He received the B.Sc. and M.Sc. in Computer Engineering, and the Ph.D. in Electronic, Computer Science and Electric Engineering from the University of Pavia, Italy, in 2003, 2005, and 2009, respectively. As a Ph.D. student he held a visiting position at the Department of Automation and Systems Engineering, University of Seville, Spain. From January 2009 to December 2010 he was a postdoctoral fellow in the Automatic Control Laboratory, ETH Zürich, Switzerland. From March 2012 to June 2012 and from August 2013 to September 2013 he was visiting scholar in Prof. Braatz Group, Department of Chemical Engineering, MIT, USA. Since December 2010 he is Assistant Professor at University of Pavia, Italy. He is the author or coauthor of more than 50 papers published in refereed journals, edited books, and refereed conference proceedings. His current research interests include control of nonlinear constrained systems, robust control, networked control, fault-tolerant control, autonomous surveillance and control of glycaemia in diabetic patients. Dr. Raimondo co-organized the NMPC Workshop on Assessment and Future Direction, Pavia, Italy, in 2008. He was also co-organizer of the invited sessions Nonlinear Model Predictive Control, for IFAC NOLCOS 2010, and New Developments in Nonlinear Model Predictive Control, for IFAC NOLCOS 2007. He served as an Associate Editor of IFAC NOLCOS 2010 and as member of the Nonlinear Model Predictive Control 2012 (NMPC'12) and European Control Conference 2013 (ECC'13) International Program Committees.
22.05.14 Building Applications for the Cloud: The Wrong, the Right, the Solid Way
Speaker: Frank Leymann, University of Stuttgart, Germany.
Date: 22.5.2014, 15:30
Since the advent of Cloud Computing Technology the question of how to use the cloud for moving existing applications to the cloud or how to build new applications for the cloud gets more and more pressing. In our talk we describe the major pitfall to avoid, namely simply bursting virtual images of stacks of applications to the cloud. Next, we describe the TOSCA standard that is targeted towards applications that are portable across multiple cloud environments, and that fosters application structures that make use of cloud benefits. Finally, we motivate the uses of best practices (“patterns”) of how to architect applications that fit best into a cloud environment.
Frank Leymann is a full professor of computer science and director of the Institute of Architecture of Application Systems (IAAS) at the University of Stuttgart, Germany. His research interests include service-oriented architectures and associated middleware, workflow- and business process management, cloud computing and associated systems management aspects, and patterns. Before accepting the professor position at University of Stuttgart he worked for two decades as an IBM Distinguished Engineer where he was member of a small team that was in charge of the architecture of IBM’s complete middleware stack.
19.05.14 Active Fault Diagnosis for Uncertain Systems
Speaker: Davide Raimondo, University of Pavia, Italy.
Date: 19.05.2014, 16:30
Component malfunctions and other faults pose a significant threat to the safety and efficiency of complex systems, such as aircrafts, power systems, and chemical plants. For these reasons, many methods have been developed to determine whether or not a fault has occured (fault detection) and, if so, the nature of the fault (fault diagnosis). The vast majority of these methods are passive, meaning that the fault status is decided on the basis of input-output data acquired during normal control operation. However, faults are not always detectable at normal operating conditions, or are obscured by the action of the control system itself. Active fault diagnosis involves injecting a suitably designed input into the system to improve the detectability and diagnosability of potential faults. The use of such inputs has clear potential for overcoming a central difficulty in fault detection, which is to distinguish the effects of faults from those of disturbances, process uncertainties, etc. This presentation discusses new methods for computing active inputs that guarantee that the input-output data of a process will be sufficient to correctly identify a fault from a given library of possible faults. To address this problem, a new formulation is considered, along with related approximations, that is amenable to efficient solution using standard optimization packages (e.g. CPLEX). The theoretical contributions combine ideas from reachability analysis, set-based computations, and optimization theory to exploit detailed problem structure and thereby manage the problem complexity.
Davide M. Raimondo was born in Pavia, Italy, in 1981. He received the B.Sc. and M.Sc. in Computer Engineering, and the Ph.D. in Electronic, Computer Science and Electric Engineering from the University of Pavia, Italy, in 2003, 2005, and 2009, respectively. As a Ph.D. student he held a visiting position at the Department of Automation and Systems Engineering, University of Seville, Spain. From January 2009 to December 2010 he was a postdoctoral fellow in the Automatic Control Laboratory, ETH Zürich, Switzerland. From March 2012 to June 2012 and from August 2013 to September 2013 he was visiting scholar in Prof. Braatz Group, Department of Chemical Engineering, MIT, USA. Since December 2010 he is Assistant Professor at University of Pavia, Italy.
He is the author or co-author of more than 50 papers published in refereed journals, edited books, and refereed conference proceedings. His current research interests include control of nonlinear constrained systems, robust control, networked control, fault-tolerant control, autonomous surveillance and control of glycaemia in diabetic patients.
Dr. Raimondo co-organized the NMPC Workshop on Assessment and Future Direction, Pavia, Italy, in 2008. He was also co-organizer of the invited sessions Nonlinear Model Predictive Control, for IFAC NOLCOS 2010, and New Developments in Nonlinear Model Predictive Control, for IFAC NOLCOS 2007. He served as an Associate Editor of IFAC NOLCOS 2010 and as member of the Nonlinear Model Predictive Control 2012 (NMPC'12) and European Control Conference 2013 (ECC'13) and 2014 (ECC'14) International Program Committees.
15.05.14 There and Back Again. Outlier Detection between Statistical Reasoning and Efficient Database Methods
Speaker: Arthur Zimek, LMU München, Germany.
Date: 15.5.2014, 17:00
Place: HS 6 (Karlsplatz)
While classical statistical methods for outlier detection had a focus on probabilistic reasoning, research on outlier detection in the database context during the last decade focused on the development of ever more efficient methods to compute outlier scores without much reasoning about the meaning of these scores. In this talk, we sketch this development and introduce some methods that go back again to statistical reasoning on top of the efficient database techniques. As we demonstrate, this opens up new possibilities for the design of ensemble methods for outlier detection.
Dr. Arthur Zimek is a Privatdozent in the database systems and data mining group of Hans-Peter Kriegel at the Ludwig-Maximilians-Universität München, Germany. 2012--2013 he was a postdoctoral fellow in the department for Computing Science at the University of Alberta, Edmonton, Canada. He finished his Ph.D. thesis in informatics on "Correlation Clustering'' in summer 2008, and received for this work the ``SIGKDD Doctoral Dissertation Award (runner-up)'' in 2009. Together with his co-authors, he received the ``Best Paper Honorable Mention Award'' at SDM 2008 and the ``Best Demonstration Paper Award'' at SSTD 2011. Zimek has served in program committees (e.g. SIGKDD, ECMLPKDD, CIKM) and as reviewer for journals like ACM TKDD, IEEE TKDE, Data Min.~Knowl.~Disc., Machine Learning etc.
06.05.14 Modeling and Solving The Patient Admission Scheduling Problem
Speaker: Andrea Schaerf, University of Udine, Italy.
Date: 6.5.2014, 17:30
The Patient Admission Scheduling problem consists in scheduling patients within a planning horizon into hospital rooms in such a way to maximize medical treatment effectiveness, management efficiency, and patient's comfort.
In this talk, we discuss, revisit, and extend the Patient Admission Scheduling problem proposed in the literature, in order to make it suitable for practical applications. We propose a solution approach based on local search, which explores the search space using a composite neighborhood. In addition, we discuss an instance generator that uses real-world data and statistical distributions so as to synthesize realistic and challenging case studies. Finally, we discuss the experimental activity including statistically-principled parameter tuning methods, for both the static and the dynamic version of the problem.
Andrea Schaerf received his master degree cum laude in Electrical Engineering in 1990 and his PhD in Computer Science in 1994 from University of Rome "La Sapienza". During his studies, he spent one year at Stanford University (California, U.S.A.) under the supervision of Y. Shoham. After graduating, he was supported for one year by a scholarship from CNR, and successively he spent one year at CWI in Amsterdam under the supervision of K. Apt, supported by a fellowship by ERCIM. From 1996 to 1998 he has been Assistant Professor at University of Rome "La Sapienza". From 1998 to 2005 he has been Associate Professor at University of Udine, where, starting November 2005, he is Full Professor. His main research interest is in algorithms, specification languages, and software tools for scheduling and timetabling problems.
30.04.14 User Mobility Patterns: A Gold Mine for Intrusion Detection of Mobile Devices
Speaker: Peter Scheuermann, Northwestern University, Evanston, Illinois, USA.
Date: 30.4.2014, 14:30
Theft of confidential data by unauthorized users accounts for much of the financial losses due to computer crime. However, most user authentication techniques for portable computers require explicit authentication. In the first part of this talk, we present an implicit re-authentication schemes for portable devices that monitors the user specific patterns based on file system and network activities. The various parameters relevant to these activities are mapped into a multidimensional vector space and a k-means clustering algorithm is used to construct a normal user model. Intruders are detected by measuring the distance between two distribution vectors.
While the above activities are useful for some types of devices, smart phones enable us to track the user behavior by observing spatio-temporal patterns. In the second part of our talk we present two statistical profiling techniques that allow us to detect anomalies corresponding to intruder patterns. Our techniques fall in the category of non-parametric collective detection anomaly models. Probabilities are extracted from the traces and anomalies correspond to a low probability region of the stochastic model. The first technique takes into account the location-in –time of users and computes the cumulative probabilities of trace samples. The second, considers also the transition probabilities between locations to construct the Markov sequence probabilities of trace samples. We present the results of our experimental evaluation based on the Reality Mining and Geolife data sets that show that our system is capable of detecting a potential intruder with 90-95% accuracy.
Peter Scheuermann is a Professor of Electrical Engineering and Computer Science at Northwestern University. He has held visiting professor positions with the Free University of Amsterdam, the Technical University of Berlin, the Swiss Federal Institute of Technology, Zurich and University of Melbourne. Dr. Scheuermann has served on the editorial board of the Communications of ACM, The VLDB Journal, IEEE Transactions on Knowledge and Data Engineering and is currently an associate editor of Data and Knowledge Engineering, Wireless Networks and ACM Transactions on Spatial Algorithms and Systems (TSAS). Among his professional activities, he has served as General chair of the ACM-SIGMOD Conference in 1988, General chair of the ER ‘2003 Conference and more recently as Program Co-Chair of the ACM-SIGPATIAL conference in 2009. His research interests are in spatio-temporal databases, mobile computing, sensor networks, data warehousing and data mining. He has published more than 140 journal and conference papers. His research has been funded by NSF, NASA, HP, Northrop Grumman, and BEA, among others. Peter Scheuermann is a Fellow of IEEE and AAAS,