DAGM 2006

28th Annual Symposium of the German Association for Pattern Recognition

  Home

  Presenter Information

  Topics

  Comittees

  Tutorials

  Important Dates

  Guest Speaker

  Registration

  Local Information

  Program

  Social Events

  Awards & Sponsors

  Picture Gallery

  Previous Conferences

  Contact

GUEST SPEAKERS

Brain-Computer Interfaces (BCI): Machine Reading of Thought-Related EEG Patterns · Gabriel Curio, Charité - Bernstein Center for Computational Neuroscience, Berlin, Germany

Macroscopic brain signals, such as the noninvasively measured scalp EEG, correlate with mental states, such as movement intentions. This opens the possibility to provide paralyzed patients, e.g., quadriplegics, with a new communication channel as their thoughts can be ‘read' by computers and utilised for triggering technical devices, such as spellers, robotic arms or wheelchairs. Leitmotiv of the Berlin BCI (BBCI) is ‘Let the machines learn', i.e., not the user is trained to produce predefined EEG patterns with command value for a computer, rather the machine learns to identify brain (EEG) signatures characteristic for naturally occurring, distinct mental states. In this overview lecture I will provide an intuitive introduction on the relevant brain anatomy and physiology, describe typical procedures for obtaining and analysing multichannel EEG data, and present a survey of recent feedback scenarios which demonstrate both, the promising scope and present limits of non-invasive BCIs.

Gabriel Curio received the Dr. med. degree with a thesis on attentional influences on smooth pursuit eye movements and holds broad specializations in Neurology and Psychiatry. Since 1991 he is leading the Neurophysics Group at the Department of Neurology of the Campus Benjamin Franklin, Charite - University Medicine Berlin . His main interest is to integrate the neurophysics of non-invasive electromagnetic brain monitoring with both basic and clinical neuroscience concepts. Recent research interests of Dr. Curio include spike-like acitivities in somatosensory evoked brain responses, neuromagnetic detection of injury currents, magnetoneurography, the comparison of cortical processing of phonems versus musical chords, speech-hearing interactions, single-trial EEG/MEG analysis and brain-computer interfacing. Since 1998 Dr. Curio serves as member of the Technical Commission of the German Society for Clinical Neurophysiology, and since 2003 as Section Editor for the IFCN journal 'Clinical Neurophysiology'.

From Bits and Bytes to Information and Knowledge: Unsupervised Learning Methods for Intelligent Information Access · Thomas Hofmann, TU Darmstadt, Darmstadt , Germany

Unstructured data is a valuable source of information and implicit knowledge. Yet, the bits and bytes of, e.g., text documents, images, or click-stream data need to be interpreted in order to transform them into business intelligence and actionable information. Clearly, this process needs to be automated to the largest possible extend in order to be scalable to the typical volumes of data. One way to accomplish this is through the use of machine learning and statistical modeling techniques. This talk will focus on unsupervised learning techniques, in particular recent developments in latent class models, dimension reduction, and matrix factorization techniques such a non-negative matrix factorization, probabilistic latent semantic analysis, latent Dirichlet allocation and related models. In addition to the methodological foundations, the talk will also present a spectrum of applications, including information search, collaborative filtering, and automatic image annotation. 

Thomas Hofmann received a Ph.D. in Computer Science from the University of Bonn in 1997 and subsequently held postdoctoral positions at the Massachusetts Institute of Technology as well as the University of California at Berkeley and the International Computer Science Institute. In 1999 he joined the Computer Science Department at Brown University as an Assistant Professor and was promoted to Associate Professor in 2004. Between 2004 and 2006, he held a position as a Professor of Computer Science at the Technical University of Darmstadt, while also serving as the Director of the Fraunhofer Institute for Integrated Publication and Information Systems. He is also co-founder and former Chief Scientist of Recommind Inc, a privately owned company focusing on enterprise search. Since July 2006, Thomas is a Director of Engineering at Google and of one the site leads of Google's engineering center in Zurich , Switzerland.

Image Processing: Interconnections · Thomas S. Huang, Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana-Champaign , Illinois , USA

We live in a complex world. To solve complex problems, we need to combine knowledge and technologies from diverse fields. Image Processing often plays a key role in these interdisciplinary problems. In a narrow sense, Image Processing comprises three areas: Coding, enhancement / restoration / reconstruction, and analysis (mensuration / detection / recognition). These three areas are of course intiminately related to each other. Many 2D images are perspective views of 3D objects and scenes. When we try to relate a 2D image to its originating 3D objects/scene, we enter the realm of Computer Vision. Recently Computer Vision techniques are increasingly used in Computer Graphics and animation. One may take the position that in a broad sense, Image Processing subsumes Computer Vision and Computer Graphics. Finally, to solve many important problems, it may be advantageous or necessary to use multimodal (esp. audio and visual) information. In this talk, we shall give two examples of interconnections. First: Very low bit-rate video coding using a 3D model-based approach, which combines Computer Vision and Computer Graphics. Second: Audio-visual speech recognition, which combines the audio and the visual modalities.

Thomas S. Huang received his B.S. Degree in Electrical Engineering from National Taiwan University , Taipei , Taiwan , China ; and his M.S. and Sc.D. Degrees in Electrical Engineering from the Massachusetts Institute of Technology, Cambridge , Massachusetts . He was on the Faculty of the Department of Electrical Engineering at MIT from 1963 to 1973; and on the Faculty of the School of Electrical Engineering and Director of its Laboratory for Information and Signal Processing at Purdue University from 1973 to 1980. In 1980, he joined the University of Illinois at Urbana-Champaign, where he is now William L. Everitt Distinguished Professor of Electrical and Computer Engineering, and Research Professor at the Coordinated Science Laboratory, and Head of the Image Formation and Processing Group at the Beckman Institute for Advanced Science and Technology and Co-Chair of the Institute's major research theme Human Computer Intelligent Interaction.

Dr. Huang's professional interests lie in the broad area of information technology, especially the transmission and processing of multidimensional signals. He has published 20 books, and over 500 papers in Network Theory, Digital Filtering, Image Processing, and Computer Vision. He is a Member of the National Academy of Engineering; a Foreign Member of the Chinese Academies of Engineering and Sciences; and a Fellow of the International Association of Pattern Recognition, IEEE, and the Optical Society of American; and has received a Guggenheim Fellowship, an A.V. Humboldt Foundation Senior U.S. Scientist Award, and a Fellowship from the Japan Association for the Promotion of Science. He received the IEEE Signal Processing Society's Technical Achievement Award in 1987 and the Society Award in 1991. He was awarded the IEEE Third Millennium Medal in 2000. Also in 2000, he received the Honda Lifetime Achievement Award for "contributions to motion analysis". In 2001, he received the IEEE Jack S. Kilby Medal. In 2002, he received the King-Sun Fu Prize, International Association of Pattern Recognition; and the Pan Wen-Yuan Outstanding Research Award. In 2005, he received the Okawa Prize. In 2006, he was named by IS&T and SPIE as the Electronic Imaging Scientist of the year. He is a Founding Editor of the International Journal Computer Vision, Graphics, and Image Processing; and Editor of the Springer Series in Information Sciences, published by Springer Verlag.

Winning the DARPA Grand Challenge · Sebastian Thrun, Artificial Intelligence Lab - Stanford University Stanford , USA

The DARPA Grand Challenge has been the most significant challenge to the robotics community in more than a decade. It required building an autonomous robot capable of traversing 132 miles of punishing desert terrain in less than 10 hours. In 2004, the best robot only made 7.3 miles. In 2005, Stanford won the challenge and the $2M prize money by successfully traversing the course in less than 7 hours. This talk, delivered by the leader of the Stanford Racing Team, will provide insights in the software architecture of Stanford's winning robot. The robot massively relied on machine learning and probabilistic modeling for sensor interpretation and control. The speaker will explain some of the basic algorithms that made this victory possible, and share some of the excitement characterizing this historic event.

Prof. Dr. Sebastian Thrun is Director of the Stanford Artificial Intelligence Laboratory (SAIL) at Stanford University . Thrun led the team that won the DARPA Grand Challenge. He also published nine books, 300 refereed papers, won numerous best paper awards, and was elected as AAAI and ECCAI Fellow. Thrun's research focuses on robotics, machine learning, and artificial intelligence.

Strategies to Avoid Segmentation · Patrice Simard, Document Processing & Understanding (DPU) group - Microsoft Research, Redmond, USA

How do differentiate Humans from Computers? This is the focus of CAPTCHAs, a Completely Automated Public Turing test to tell Computer and Human Apart. From a pattern recognition standpoint, CAPTCHAs are interesting because they can be split into a segmentation challenge and a recognition challenge. We postulate that the (pure) recognition challenges are easy. Indeed, if the segmentation part of a CAPTCHA

is solved, the state of the art recognition algorithms beat humans on every CATCHA that we tried. However humans remained far superior at segmentation. This has inspired us to do “segmentation-less” algorithms for handwriting recognition. The methods are applicable in other fields such as vision.

Patrice Simard is Chief Scientist in Microsoft Live Labs. His team is responsible for developing new technologies related to the internet. Examples include search, entity extraction, spam, images processing, mobile computing, and document analysis. Prior to Live Labs, Patrice was Research Area Manager at Microsoft Research. From 2002 to 2005, he also managed the Document Processing and Understanding group at Microsoft Research.

Since he joined Microsoft in 1998, Patrice has developed and shipped algorithms for compression, handwriting recognition, and document analysis. Patrice has filed over 50 patents (awarded or pending) and is the author of over 40 scientific publications in the fields of Machine learning, Vision, and Signal Processing. Before joining Microsoft, Patrice worked in research with AT&T Laboratories from 1991 to 1998. He received his bachelor's degree in Electrical Engineering (1986) from l'Université de Montréal, Canada and his Ph.D. in Computer Science (1991) from the University of Rochester, NY. His scientific interests include algorithms, compression, learning, and generalization.