Department of Informatics

News
Department
Teaching
Research
Agenda
Akkreditierungen
Archive  

Details Colloquium Fall 2009

24.9.2009 - Efficiently Rendering Realistic Images

Speaker: Prof. Matthias Zwicker
Host: Renato Pajarola, IFI, UZH

Abstract

Developing algorithms to render realistic images is one of the core topics in computer graphics. Applications for realistic rendering technology range from the digital entertainment industry to computer aided industrial design to architecture or archeology. One of the main challenges in realistic rendering is that it requires a large amount of computation. Even on todays hardware, it may still take minutes to hours to compute a single image using one CPU. In this talk I will explain why popular methods for realistic image synthesis are so computationally expensive. I will then present our recent research that focuses on making these computations more efficient. I will also describe how it is possible to render realistic images in mere fractions of a second using pre-computation.

Bio

Matthias Zwicker is a professor at the University of Bern and the head of the Computer Graphics Group at the Institute for Computer Science and Applied Mathematics. From 2006 to 2008 he was an Assistant Professor with the Computer Graphics Laboratory at the University of California in San Diego. He obtained a PhD from ETH Zurich and was a post-doctoral associate with the computer graphics group at the Massachusetts Institute of Technology for two years. Prof. Zwicker's research focus is in computer graphics. He is most interested in high-quality rendering, signal processing for image synthesis, point-based methods for rendering and modeling, and data-driven modeling and animation.

8.10.2009 - Knowledge Management Success Factors - Experiences from over a Decade of Successful KM Initiatives

Speaker: Dipl.-Ing. Frank Leistner
Host: Michael Hess, IFI, UZH

Abstract

Knowledge Management (KM) is a discipline that is not always well defined and often seen as a primarily technical discipline. Through 12 years of KM initiatives within the globally distributed Software vendor SAS the key success factors have proven to be a lot more in the non-technical areas. Elements like motivation, marketing, evangelism and knowledge intermediary support have had more impact on the success than the use of specific technologies. Together with the more common understood elements of KM those soft factors form more of a long-term initiative than a project. The presentation will cover several case studies from the areas: 1. Technical Knowledge Exchange 2. Skills and Resource Management 3. Collaborative KM. As a conclusion common key success factors, lessons learned and insights into ways to deal with measuring the impact of KM will be discussed.

Bio

Frank Leistner is responsible for driving internal Knowledge Management within SAS via worldwide initiatives. He has been with the company for 16 years, formerly working for Siemens-Nixdorf in a US-Germany liaison role. Mr. Leistner holds an M.S. in Computer Science from SUNY Albany, NY, USA and a Dipl. Inf. (Computer Science) degree from Technical University Carolo-Wilhemina, Braunschweig, Germany. He has been guiding worldwide knowledge management projects within SAS since 1997 and worked with the IBM Institute for Knowledge Management from 1999-2003. He is one of the authors of the book "Leading with Knowledge, Knowledge Management Practices in Global Infotech Companies", Madanmohan Rao, 2003. He has been a member of the Harvard Learning Innovations Laboratory round-table since 2003 and the Babson Working Knowledge Program since 2006, invited by President's Distinguished Professor in Information Technology and Management at Babson College Thomas H. Davenport. Frank's hobbies include juggling and motorcycle riding.

22.10.2009 - Why are Software Engineers a Pain in the Neck?

Speaker: Prof. Mehdi Jazayeri
Host: Harald Gall, IFI UZH

Abstract

Software is critical to the working of our modern society. Software is pervasive, it works, it is invisible, it just seems to run things well and stay out of the way. This is not by accident. Over time, software engineers have developed techniques for creating functioning software that is efficient and dependable. Software is so smoothly woven into our machines and services that despite its pervasiveness, most people do not even realize its presence in the devices and services they use.Software engineering is about understanding problems and implementing solutions that will work now and forever, economically, reliably, and efficiently. Software engineers have developed an approach and a way of thinking to tackle problems and look for near-perfect solutions. Unfortunately, this attitude does not make them popular with managers, who would prefer a less perfect solution that is delivered on time and at less cost. The approach does not make them popular with other computer scientists (e.g. computational scientists) who just want to get the software running and get the results out, never mind guarantees of correctness or adaptability to future needs. Worse, looking for perfection does not work well in the real world with lay people. Most people are happy to leave things as they are as long as they sort of work. They don't need software engineers to point out all the existing or potential bugs that could be fixed to make processes more efficient and general.In this talk, I will discuss some fundamental principles of software engineering that are crucial to producing good software. The principles are more general and can help for all problem solving. Unfortunately, applying them in contexts where the software engineering culture is not understood can make you a pain the neck. I hope the talk will be enlightening to non-software engineers and at least entertaining for software engineers.

Bio

Mehdi Jazayeri is professor of computer science and founding dean of the Faculty of Informatics at the University of Lugano since October 2004. Before that he was a professor and head of the Distributed Systems Group at the Technical University of Vienna (1994-2004). He worked at several startup companies in Silicon Valley before joining Hewlett-Packard Laboratories in Palo Alto for ten years (1984-94). He began his career as an assistant professor at the Computer Science Department of the University of North Carolina at Chapel Hill (1975-1980). Mehdi Jazayeri is an IEEE Fellow and was program co-chair of ICSE 2000 and program chair of ESEC-FSE 1997.

29.10.2009 - The Spec# Programming System

Speaker: Prof. Peter Müller
Host: Martin Glinz, IFI, UZH

Abstract

Spec# is an experimental programming system that seeks to simplify the development of correct programs. The system includes the object-oriented programming language Spec#, which is a superset of the C# language, adding type features like non-null types and contract features like pre- and postconditions and object invariants. The contracts are checked dynamically through compiler-inserted run-time checks, and they can also be checked statically using the Spec# automatic program verifier. Spec# is being developed by Microsoft Research in close collaboration with ETH Zurich.

Bio

Peter Müller is a Full Professor and head of the Chair of Programming Methodology at ETH Zurich since August 2008. His research focuses on languages, techniques, and tools for the development of correct software. His previous appointments include a position as Researcher at Microsoft Research in Redmond, an Assistant Professorship at ETH Zurich, and a position as Project Manager at Deutsche Bank in Frankfurt. Peter Müller received his PhD from the University of Hagen.

12.11.2009 - Description Logic Based Ontology Languages

Speaker: Prof. Ian Horrocks
Host: Avi Bernstein, IFI, UZH

Abstract

Description Logics (DLs) are a family of class based knowledge representation formalisms characterized by the use of various constructors to build complex classes from simpler ones, andby an emphasis on the provision of sound, complete and (empirically) tractable reasoning services. Although they have a range of applications (e.g., configuration and information integration), they are perhaps best known as the basis for widely used ontology languages such as OWL. The decision to base OWL on DLs was motivated by a requirement that key inference problems be decidable, and that it should be possible to provide "practical" reasoning services to support ontology design and deployment. This talk will introduce description logics and their relationship to OWL, and then describe some of the algorithms and implementation techniques that have enabled them to satisfy this requirement in spite of the discouraging worst case complexity of the relevant decision problems.

Bio

Ian Horrocks is a Professor in the Oxford University Computing Laboratory. He is internationally recognized for his work on ontology languages and reasoning systems. He was centrally involved in the development of the OIL, DAML+OIL and OWL ontology languages, and is co-chair of the W3C Working Group that is developing OWL 2. He has published more than 150 articles in conferences, journals and books. In 2005 he received the BCS Roger Needham award for his his pioneering work in the Semantic Web and was also awarded an EPSRC Senior Research Fellowship.

26.11.2009 - Reputation-based Trust in Peer-to-peer Communities

Speaker: Prof. Karl Aberer
Host: Burkhard Stiller, IFI, UZH

Abstract

Trust is a key problem in peer-to-peer systems where interactions are performed usually with complete strangers. Reputation-based trust has been proposed and deployed as a solution to deal with this problem. In this talk we will first review basic mechanisms of reputation-based trust mechanisms based on statistical approaches. We then present solutions for tackling some key problems with these approaches. We investigate the effect of reputation-based trust management on strategically acting agents. It turns out that the threat of punishment fosters cooperation at very tolerable cost. However, stable identification is a weak spot of all reputation-based trust mechanisms. We propose a novel mechanism based on a forward feedback mechanism for reliable service provisioning in peer-to-peer networks that requires only mutual identification among direct neighbors and thus avoids global identification. Finally we give an outlook on how recent developments on the semantic and social Web are leading to new applications and opportunities for reputation-based trust.

Bio

Karl Aberer is a Professor for Distributed Information Systems at EPFL Lausanne, Switzerland, and director of the Swiss National Centre for Mobile Information and Communication Systems (NCCR-MICS). His research interests are on decentralization and self-organization in information systems with applications in peer-to-peer search, overlay networks, trust management and mobile and sensor networks. Before joining EPFL in 2000 he was leading the research division of open adaptive information systems at the Integrated Publication and Information Systems Institute (IPSI) of GMD in Germany, which he joined in 1992. There his work concentrated on XML data management and cross-organizational workflows. He studied mathematics at ETH Zürich where he also completed his Ph.D. in theoretical computer science in 1991. From 1991 to 1992 he was postdoctoral fellow at the International Computer Science Institute (ICSI) at the University of California, Berkeley. He is member of several journal editorial boards, including VLDB Journal, and conference steering committees. Recently he served as PC co-chair of ICDE 2005, MDM 2006 and ISWC 2007.