May 24, 2005 - Dr. Verena Hafner: Robots: Tools or Toys? Some Answers from Biorobotics, Developmental and Entertainment Robotics

• Dr. Verena V. Hafner, Sony Computer Science Laboratory, Paris, France
• Dr. Verena V. Hafner is a researcher in Computer Science, AI and Robotics. She received her M.Res. in Computer Science and Artificial Intelligence with distinction from the University of Sussex (UK) in 1999, after completing her Undergraduate studies in Mathematics and Computer Science in Germany. In 2004, she received her Ph.D. in Natural Sciences from the University of Zurich, Switzerland, where she was assistant researcher at the Artificial Intelligence Lab. From 2004 till 2005, she worked as an associate researcher in the developmental robotics group at Sony CSL in Paris, France. Her research interests include neural computation and spatial cognition in thearea of bio-robotics, and developmental robotics with a focus on joint attention, communication and motivation. A list of publications can be found: http://www.verena-hafner.de/
• Summary: Robots will play an increasingly important role in the development of Artificial Intelligence. Many researchers are convinced that for machines to become truly intelligent, they will need to be embodied. Contemporary research in intelligent robotics does not only draw inspiration from biology and psychology, but also provides thesedisciplines with the means to test hypotheses by using robots as research tools. Outside the lab, the main market today for robots is moving away from industry robots to the home, with robotic companions being developed for entertainment purposes. In this talk, several examples of cutting-edge research in robotics will be presented, and their implications for the future of robotics will be discussed.

May 31, 2005 - Dr. Justus Piater: Computer Vision

• Prof. Dr. Justus Piater, University of Liège, Belgium
• After graduating with highest honors from the University of Magdeburg, Germany, Justus H. Piater was awarded a Fulbright scholarship and obtained his Ph.D. in computer science at the University of Massachusetts Amherst, USA, in 2001. A recipient of a European Marie-Curie Individual Fellowship, he was a postdoctoral researcher at the project PRIMA, INRIA Rhône-Alpes, France, from 2000 to 2002. Since 2002 he is an assistant professor of computer science at the University of Liège, Belgium, where he directs the Computer Vision Research Group. His research interests include computer vision and Machine learning, witha focus on biologically-inspired visual learning and interactive vision, and video analysis.
• Summary: Computer Vision is a multi-faceted field that is driven by a variety of applications in various fields such as car license-plate reading medical image analysis, image retrieval, robot navigation, or movie special effects. The talk will consist of three parts, the first of which illustrates some of the major applications and objectives of computer vision research. The second part gives a gentle introduction to some of the key underlying techniques. During the third part, I will give an overview of current research in my group on video analysis and visual learning.

June 7, 2005 - Dr. Thomas Laengle: Multi-Agent-Approach for Control and Diagnosis of Complex Production Systems

• Dr. Thomas Längle, University of Karlsruhe and Fraunhofer Institute, Karlsruhe, Germany
• Dr. Thomas Längle studied Computer Science at the University of Karlsruhe in Germany, where he received his Master of Science in 1993 with a work on the localization of robot systems. Then, he moved to the Institute for Real-Time Computer Systems and Robotics at the University of Karlsruhe, where he worked as a researcher in the robotics group. Since 1995, he is leader of the research group Intelligent Robots, he received his PhD in December, 1996. In January 1997, he moved to the Institute for Process Control and Robotics at the University of Karlsruhe, where he worked as leader of the research group Intelligent Control and Diagnosis. The group consisted of ten full-time researchers, PhD-students and several master students. In December 2003, he finished his habilitation in computer science. His research interests include different aspects of robotics and manufacturing, intelligent control, multi-agent-systems, diagnosis and tele-service.
• Summary: The main advantage of distributed controlled systems in factory automation is the decentralized task execution by the system components. This way, properties for the design of flexible control architectures like modularity, fault-tolerance, integrative and extendibility are easy to obtain, further it is possible to use the concepts of distributed knowledge and decentralized world representation. On the other hand, a robust diagnosis of the overall systems is very difficult to guarantee due to changing system configuration. Further problems occur by moving from the non-embedded to the embedded world. To overcome these problems and to realize a modular and flexible system, a multi-agent architecture is proposed. Further, methods of semantically and spatially distributed diagnosis as a means of agent co-operation are introduced that can also be used for embedded systems.

June 14, 2005 - Dr. Toni Bollinger: Data Mining

• Dr. Toni Bollinger, IBM Research & Development Böblingen, Germany
• Dr. Toni Bollinger studied Computer Science in Bonn, Germany, and Paris. He joined IBM in 1987 where he worked in several AI research projects. Since 1994 he works on data mining in the development of the Intelligent Miner software products as well as in practical data mining projects. His main focus in his current activities is to make data mining easier and accessible to a wider range of users.
• Summary: Knowledge Discovery and Data Mining deal with finding valuable information in huge volumes of data. They are well established disciplines not only in the academic world, but also in the industry where commercial systems are available frommajor software vendors. This success is in particular due to the interdisciplinary nature of the field having its origins in Artifical Intelligence, notably Machine Learning, but also Statistics and Database Research. The talk will give an overview of the field. It will present some of the major data mining techniques together with its application areas and solutions that are built upon it. The talk will conclude with a discussion of current trends and challenges of data mining.

June 21, 2005 - Dr. Paul Buitelaar: Language Technology in Knowledge Markup and Ontology Evolution

• Dr. Paul Buitelaar, German Research Center for Artificial Intelligence (DFKI), Saarbrücken, Germany
• Dr. Paul Buitelaarstudied Computational Linguistics at Utrecht University, the Netherlands,and Computer Science at Brandeis University, Waltham, MA, USA (PhD 1998). He organized several international workshops on topics in lexical semantics, semantic annotation,ontology development and semantic web and has been an invited speaker at many panels and workshops in these areas. His main research interest is in Language Technology for Semantic-Based Information Access. At the DFKI he has been working as a senior researcher and project leader on a number of EUand German funded projects (e.g. currently the project SmartWebon Mobile Broadband Access to the Semantic Web) and as a project coordinator of the EU/NSF funded project MuchMoreon concept-based cross-lingual medical information retrieval (2000-2003). He was initiator and coordinator of the Special Interest Group on Language Technology inOntology Development and Use within the EU funded thematic network OntoWeb (2001-2004) and is co-chair of the DFKI Competence Center Semantic Web.
• Summary: In this talk I will present an overview of Human Language Technology (HLT) and its use in Semantic Web development. HLT is concerned with automatic linguistic processing towards the semantic analysis and extraction of information from textual data. In the context of the Semantic Web the use of HLT is in knowledge markup of web documents for ontology population and text mining for ontology evolution (extension andmodification of ontology models). The talk will include examples of both as currently developed in the context of the SmartWebproject on "Mobile Broadband Access to the Semantic Web" http://www.smartweb-projekt.de/

June 28, 2005 - Dr. René Witte: Text Mining

• Dr. RenéWitteUniversity of Karlsruhe, Germany and CLaC Lab, Concordia University, Montréal, Canada.
• Dr. RenéWitte studied Computer Science at the University of Karlsruhe,Germany, where he received his Diploma in 1996. In July 2002 he received his Dr.-Ing.from the Faculty of Informatics at the Karlsruhe University,as a member of Prof. Dr. Peter C. Lockemann'sInformation Systems Group at the Institute for Program Structures and Data Organization (IPD). He also spent considerable time as a freelancing industry consultant inthe area of information system engineering. Currently, he splits his time between research work at the CLaC Laboratoryat Concordia University in Montréal, Québec, Canada, where he works on language technology foundations and their applications with in the Fungal Web bioinformatics project and the Information Systems Group at the University of Karlsruhe, where he works on a DFG-funded project analysinga 19th century encyclopedia of architecture in cooperation with architects, building historians, and computer scientists.

• Summary: Do you have a lack of information? Or do you rather feel overwhelmed by the sheer amount of (online) available content, like emails, news, web pages, and electronic documents? The rather young field of Text Mining developed from the observation that most knowledge today is hidden within natural language documents and thus cannot be automatically processed by classical information systems. Text Mining is a highly interdisciplinary field, drawing on foundations and technologies from computational linguistics, database systems, and artificial intelligence, but applying these in a new and often unconventional way. In this talk, we give an introduction to the field of text mining and show some of its linguistic and technical foundations. Several application examples are examinedin detail, especially the automatic summarization of documents within the DUC competition and the extraction of biological knowledge from research papers.

July 5, 2005 - Prof. Dr. Torsten Schaub: Knowledge Representation and Reasoning via Answer Set Programming

• Prof. Dr. Torsten Schaub, University of Potsdam, Germany
• Torsten Schaub received his diploma and dissertation in informatics from the Technical University of Darmstadt, Germany, in 1990 and 1992, and his habilitation in informatics in 1995 from the University of RennesI, France. From 1990 to 1993 he was a researcher at the Technical University at Darmstadt,from 1993 to 1995 Research Associate at IRISA/INRIA in Rennes. From 1995 to 1997, he was Professor at the University of Angers where he founded the research group FLUX dealing with the automatisation of reasoning from incomplete, contradictory, and evolutiveinformation. Since 1997, he is Professor for knowledge processing and information systemsat the University of Potsdam. In 1999, he became Adjunct Professor at the School of Computing Science at Simon Fraser University. Since 2002 Torsten Schaub serves as head of the Institute of Informatics. His particular research interests range from the theoretic foundations to the practical implementation of methods for reasoning from incompleteand/or inconsistent information.
• Summary:The field of Knowledge Representation and Reasoning is central to Artificial Intelligence. Interestingly, the field has been going through a methodological shift during recent years. While the past was dominated by query-oriented reasoning, model-based techniques become more and more popular nowadays. This is accompanied by the availability of highly efficient general purpose problem solvers, among which satisfiability and Answer Set solvers are the most prominent ones. The general idea is then totranslate an application problem into a logical specification. This specification is in turn passed to a solver, which outputs models representing solutions to the initial applicationproblem. The talk will provide an introduction to Answer Set Programming (ASP), in which applications are specified in terms of sets of logical rules. ASP has its roots in deductive databases, non-monotonic reasoning and declarative programming. It is closely related to constraint satisfaction and propositional satisfiability. And it has already been proved to be an effective tool in a range of applications from planning to model checking to combinatorial and combinatorial optimization.