Οι ομιλίες παρουσιάζονται στην Αίθουσα Τηλεδιασκέψεων Κέντρου Διαχείρισης Δικτύων. Πιθανές εξαιρέσεις θα ανακοινώνονται ανά περίπτωση.

29 Μαίου (Δευτέρα, 11:00)
Ομιλητής: Prof. Anil K. Jain
"Biometric Recognition: How Do I Know Who You Are?"
 

A wide variety of systems require reliable personal recognition schemes to either confirm or determine the identity of an individual requesting their services. The purpose of such schemes is to ensure that only a legitimate user, and not anyone else, accesses the rendered services. Examples of such applications include (i) secure access to buildings, ATMs, and cellular phones, (ii) obtaining driver licenses and welfare benefits, and (iii) issuing passports and visas. Biometric recognition, or simply biometrics, refers to the automatic recognition of individuals based on their physiological and/or behavioral characteristics (e.g., fingerprint, face, iris, voice). Biometrics allows us to confirm or establish an individual's identity based on "who she is", rather than by "what she possesses" (e.g., an ID card) or "what she remembers" (e.g., a password). Biometric systems also introduce an aspect of user convenience; they alleviate the need for a user to "remember" multiple passwords associated with different applications. However, design of a biometric system is a complex and challenging pattern recognition task. The system designer has to contend with has to contend with problems related to noise in the sensed data, large intra-class variability, large inter-class similarity, invariant representation, large number of classes and threats to the biometric system itself. Some of these problems can be addressed by multimodal biometric systems that fuse the evidence presented by multiple identity traits of a user. In spite of the fact that the first Automatic Fingerprint Identification System (AFIS) was installed in 1960s, biometric recognition remains a difficult pattern recognition problem. In this talk, we will give an overview of biometrics, representation and matching algorithms for fingerprints, state-of-the-art performance figures and the associated system security and privacy issues.

Short Biography:

Anil Jain is a University Distinguished Professor in the Departments of Computer Science & Engineering, Electrical & Computer Engineering and Statistics & Probability at Michigan State University. His research interests pattern recognition, data clustering and biometric authentication. He received awards for best papers in 1987 and 1991 from the Pattern Recognition Society. He also received the 1996 IEEE Transactions on Neural Networks Outstanding Paper Award. He was the Editor-in-Chief of the IEEE Trans. Pattern Analysis and Machine Intelligence. He is a Fellow of the AAAS, ACM, IEEE, IAPR and SPIE. He has received a Fulbright Research Award, a Guggenheim fellowship, the Alexander von Humboldt Research Award and the 2003 IEEE Computer Society Technical Achievement Award. ISI has designated him as a highly cited researcher. Holder of six patents in the area of fingerprint matching, he is the author of a number of books including Handbook of Multibiometrics, Springer 2006, Handbook of Face Recognition, Springer 2005, Handbook of Fingerprint Recognition, Springer 2003, BIOMETRICS: Personal Identification in Networked Society, Kluwer 1999 and Algorithms for Clustering Data, Prentice hall, 1988. He is an Associate editor of the IEEE Transactions on Information Forensics and Security and ACM Transactions on Knowledge Discovery in Data.

13 Απριλίου (Πέμπτη, 11:00)
Ομιλητής: Prof. Ali H. Sayed
"Energy Conservation in Adaptive Filtering"
 

Adaptive filters are systems that respond to variations in their environment by adapting their internal structure in order to meet certain performance specifications. Such systems are widely used in communications, biomedical applications, signal processing, and control. The performance of an adaptive filter is evaluated in terms of its transient behavior and its steady-state behavior. The former provides information about how fast a filter learns, while the latter provides information about how well a filter learns. Such performance analyses are usually challenging since adaptive filters are, by design, time-variant, nonlinear, and stochastic systems. For this reason, it has been common in the literature to study different adaptive schemes separately due to the differences that exist in their update equations. The purpose of this talk is to provide an overview of an energy conservation approach to the performance analysis of adaptive filters. The framework is based on studying the energy flow through successive iterations of an adaptive filter and on establishing a fundamental energy conservation relation; the relation bears resemblance with Snell’s Law in optics and has far reaching consequences to the study of adaptive schemes. In this way, many new and old results can be pursued uniformly across different classes of algorithms. In particular, the talk will highlight some recently discovered phenomena pertaining to the learning ability of adaptive filters. It will be seen that adaptive filters generally learn at a rate that is better than that predicted by least-mean-squares theory; that is, they are "smarter" than originally thought! It will also be seen that adaptive filters actually have two distinct rates of convergence; they learn at a slower rate initially and at a faster rate later; perhaps in a manner that mimics the human learning process. The talk will also discuss adaptive distributed systems that are able to exploit both the temporal and spatial dimensions of the data collected at spatially distributed nodes in order to enhance the robustness of the processing tasks and improve the probability of signal and event detection. Adaptation is needed not only because the environmental conditions vary with time and space, but also because the network topology may vary.

Short Biography:

Ali H. Sayed is Professor and Chairman of Electrical Engineering at UCLA, where he also directs the Adaptive Systems Laboratory (www.ee.ucla.edu/asl). He has published widely in the areas of adaptive filtering, estimation theory, and signal processing for communications with over 250 articles and 4 books, including the textbooks Fundamentals of Adaptive Filtering (Wiley, NY, 2003) and Linear Estimation (Prentice Hall, NJ, 2000). He is a Fellow of IEEE and has served as Editor-in-Chief of the IEEE Transactions on Signal Processing (2003-2005). He serves as Editor-in-Chief of the EURASIP Journal on Applied Signal Processing. His research has received several recognitions including the 1996 IEEE Donald G. Fink Prize, 2002 Best Paper Award from the IEEE Signal Processing Society, 2003 Kuwait Prize, 2005 Frederick E. Terman Award, 2005 Young Author Best Paper Award from the IEEE Signal Processing Society, and two Best Student Paper Awards at international meetings (1999,2001). He has served as a 2005 Distinguished Lecturer of the IEEE Signal Processing Society and as a member of the Publications (2003-2005) and Awards (2005) Boards of the same society. He is a member of the Signal Processing Theory and Methods (SPTM) and Signal Processing for Communications (SPCOM) technical committees of the IEEE Signal Processing Society. He is also serving as General Chairman of ICASSP 2008.

12 Απριλίου (Τετάρτη, 12:00)
Ομιλητής: Prof. Leandros Tassiulas
"Information flow issues in cross-layer models of wireless communication networks"
 

Advances in communication technology over the last several years made possible the deployment of broadband wireless networks that provide integrated services via inexpensive low-powered mobile terminals. End users often expect seamless transition from wire-line to wireless networks. That requires quality of service provisioning that is compatible in the wireless and the wire-line parts of the network. Wireless networks though exhibit peculiarities due to which meeting stringent quality of service requirements becomes a rather challenging task. The volatile, error-prone mobile channel on one hand and the interference limited radio medium on the other, necessitate a cross-layer approach in the design of higher layers. We will present various approaches for network control at the access and network layer where the controller should rely on channel state information passed from the physical layer, while making resource allocation decisions. Furthermore several considerations belonging typically to the physical layer like channel coding rate, signal constellation selection as well as power level adjustments, frequency selection and beam steering in multiple antenna systems are to the disposal of the access controller in several current schemes for broadband wireless access. We will present approaches for dealing with these design choices and discuss related state-of-the art broadband access technologies. The impact of these techniques on the efficiency in terms of bandwidth, spectrum utilization and energy consumption will be discussed.

Short Biography:

Leandros Tassiulas is Professor in the Dept of Computer Engineering and Telecommunications at the University of Thessaly Greece since 2002. His research activity over the last fifteen years is towards the development of communication and information processing networks that facilitate access and exchange of information among multiple entities. Current research and teaching topics include wireless mobile communications, ad-hoc networks, smart antennas, sensor networks, high speed networked environments. He was Assistant Professor. at Polytechnic University, NY, 1991-1995, Associate Prof. at the University of Maryland, College Park until 2002 (on leave 2000-2002) and Professor of Computer Science at the University of Ioannina Greece 1999-2002. He obtained the Diploma in Electrical Engineering from the University of Thessaloniki, Greece in 1987, and the M.S. and Ph.D. degrees in Electrical Engineering from the University of Maryland, College Park in 1989 and 1991 respectively. He has been Associate Editor for Communication Networks for IEEE Transactions on Information Theory and an editor for IEEE/ACM Transactions on Networking. His research activity received several recognitions including a National Science Foundation (NSF) Research Initiation Award in 1992, an NSF CAREER Award in 1995, the Office of Naval Research Young Investigator Award in 1997 and the INFOCOM `94 best paper award. In 1999, he was awarded the "Bodossaki Foundation Academic Prize" in the field: Applied Science: Theories, Technologies and Applications of Parallel and Distributed Computing Systems, that is awarded

15 Μαρτίου (Τετάρτη, 12:00)
Ομιλητής: Prof. Michael Georgiopoulos
"ART Neural Network Architectures"
 

Fuzzy ARTMAP (FAM) is currently considered to be one of the premier neural network architectures in solving classification problems. Fuzzy ARTMAP is one of the most popular ART architectures, developed by Carpenter and Grossberg in 1992. Some of the advantages that Fuzzy ARTMAP possesses (as well as other ART architectures) is that it can solve arbitrarily complex classification problems, it converges quickly to a solution (within a few presentations of the list of the input/output patterns belonging to the training set), it has the ability to recognize novelty in the input patterns presented to it, it can operate in an on-line fashion (new input/output patterns can be learned by the system without re-training with the old input/output patterns), and it produces answers that can be explained with relative ease. Since its inception a number of modifications of Fuzzy ARTMAP have appeared in the literature with the ultimate goal of improving Fuzzy ARTMAP’s performance. This seminar will give a short but succinct overview of ART, and emphasize some of the ART architectural innovations that the speaker and his colleagues have developed at the University of Central Florida. In particular, some emphasis will be placed on a class of ART neural network architectures that is referred to as semi-supervised ART architectures, such as semi-supervised Fuzzy ARTMAP (FAM) and semi-supervised Ellipsoidal ARTMAP (EAM). The purpose of these architectures is to design ART networks that are of smaller size than the Fuzzy ARTMAP and Ellipsoidal ARTMAP architectures without affecting the generalization performance. Designing more compact classifiers to solve classification problems is desirable because then the designer and the user can more easily interpret the answers that the classifier provides. Furthermore, we are going to discuss one of the most recent innovations of our ART research at the University of Central Florida. This innovation is an ART structure referred to as genetically engineered ART that allows one to mix up the solutions provided by many trained ART neural networks with the objective (again) to design ART neural network classifiers of small size and good generalization. We intend to provide a thorough comparison of a number of ART neural network classifiers on a variety of benchmark simulated and real classification problems. This comparison will focus on ART classifiers, such as FAM, semi-supervised FAM (ssFAM), EAM, semi-supervised EAM (ssEAM), Gaussian ARTMAP (GAM), micro-ARTMAP, and obviously the recently introduced genetically engineered Fuzzy ARTMAP (G-FAM). This comparison will provide the attendee with a good understanding of how all these ART classifiers fair in solving classification problems. Some limited comparisons with other classifiers such as multi-layer perceptrons (MLPs) and Support Vector Machines (SVMs) will also be discussed.

Short Biography:

Dr. Georgiopoulos has actively worked on research areas such as, communication networks, spread spectrum communications, neural networks, and applications of neural networks in pattern recognition, image processing, computer vision, manufacturing, and computer generated forces modeling. He has published over 50 papers in journals and over 150 papers in conferences in the areas of neural networks, and communications. He has been a principal or co-principal investigator on research grants totaling more than $7.0M. Lockheed Martin, Harris, US Army, US Navy, DMSO, NSF, Lockheed Martin, Harris, and other agencies from the State of Florida have funded his research. Dr. Georgiopoulos is a Senior Member of IEEE and a member of the International Neural Network Society. He is Serving as an Associate Editor of the IEEE Transactions on Neural Networks since 2001, and as an Associate Editor of the Neural Networks journal since 2006.