Plenary Lectures

Phil Woodland, Cambridge, UK
Discriminative Training and Adaptation of Speech Recognition Systems

Tuesday August 26th, 8:30am - 9:30am

Yonina Eldar, Technion, Israel
Beyond Bandlimited Sampling: Nonideal Sampling, Smoothness and Sparsity

[slides in pdf]

Wednesday August 27th, 2:10pm - 3:10pm

Digital applications inherently rely on sampling a continuous-time signal to obtain a discrete-time representation. Many of the limitations encountered in current digital-to-analog converters stem from the traditional assumption that data must be acquired at the Shannon-Nyquist rate, corresponding to twice the bandwidth. A major drawback of the Shannon paradigm is that natural signals can often be better represented by models beyond bandlimited. Furthermore, ideal point-wise sampling and sinc interpolation are difficult to implement.

In this talk we present several extensions of the Shannon theorem which accommodate a broader class of input signals as well as nonideal sampling and nonlinear distortions. This framework is based on viewing sampling in a broader sense of projection onto appropriate subspaces, and then choosing the subspaces to yield interesting new possibilities such as below Nyquist sampling of sparse signals, pointwise sampling of non bandlimited signals, and perfect compensation of nonlinear effects.

We begin by presenting a broad class of sampling theorems for signals confined to an arbitrary subspace in the presence of non-ideal sampling, and nonlinear distortions. Next, we develop minimax recovery techniques that best approximate an arbitrary smooth input signal, using simple interpolation kernels. Finally, we discuss sparse analog signals that can be represented by a disjoint set of bands in some transform domain. Combining traditional sampling ideas with results from the field of compressed sensing we show how to reconstruct an analog multi-band signal from minimum-rate samples when the band locations are unknown.

Jeffrey Fessler, University of Michigan, USA
Signal processing in medical image reconstruction

[slides in pdf]

Thursday August 28th, 2:10pm - 3:10pm

Iterative methods for image reconstruction based on statistical and physical models have become available commercially for PET and SPECT systems over the past ten years. Commercial MRI and X-ray CT scanners continue to use classical (noniterative) image reconstruction methods. This talk will summarize some of the signal processing methods that have led to improved image reconstruction for medical imaging as well as some of the techniques currently under development that may offer clinical benefits in the near future, provided the computational challenges are overcome.

Claude Lamblin, France Telecom
Recent Audio/Speech Coding Developments in ITU-T and future trends

[slides in pdf]

Thursday August 28th, 5:30pm - 6:30pm

In order to provide a universal multimedia access to users, various networks and terminals are interconnected with different access technologies. To cope with this heterogeneity, audio coding development has been driven by three objectives: quality enhancement, flexibility improvement and increase of robustness. After decades of bit rate lowering, there is now a clear evolution toward more bandwidth for a higher Quality of Service. A major flexibility improvement is brought by scalable codecs capable to interoperate at various bit rates and bandwidths. Meanwhile, the work on development of additional features (Voice Activity Detector, Packet Loss Concealment procedure) necessary to accommodate current coders to new application requirements has continued. The presentation will start with a broad overview of ITU-T speech and audio coding standards. It will then focus on the 2005-2008 Study Period achievements. Finally the main orientations for next Study Period (2009-2012) will be discussed.

Bernhard Schaelkopf, Max Planck Institute, Tuebingen, Germany
Kernel Tricks, Means and Ends

[slides in pdf]

Friday August 29th, 2:10pm - 3:10pm

Support vector machines and other kernel methods have become one of the most widely used techniques in the field of machine learning. I will present my thoughts on what made them popular and what may (or may not) keep them going. In the talk, I also plan to describe several kernel algorithms, in particular kernel mean methods on homogeneity testing, and discuss applications in different domains, including computer graphics.

EURASIP Fellow Inaugural Lectures

Bjorn Ottersten, KTH - Royal Institute of Technology, Stockholm, Sweden
The evolution of mobile communications - a signal processing perspective

[slides in pdf]

Tuesday August 26th, 11:50am - 12:50pm

Pierre Duhamel, L2S/CNRS/SUPELEC, France
Cheating or being optimistic ? An illustration with Joint Source and Channel Coding/Decoding

[slides in pdf]

Wednesday August 27th, 5:30pm - 6:30pm

General Chairman

J.-Ph. Thiran, EPFL, Switzerland

General Co-Chairman
P. Vandergheynst, EPFL, Switzerland

Technical Program Co-Chairmen
P. Frossard, EPFL, Switzerland

A. Cavallaro, Queen Mary,
University of London, UK

Plenary Talks
M. Unser, EPFL, Switzerland

S. Godsill, Cambridge Univ, UK

Special Sessions
C. De Vleeschouwer, UCL, Belgium

J. Louveaux, UCL, Belgium

N. Paragios, Ecole Centrale, France

F. Bimbot, IRISA, Rennes, France

R. Reilly, UC Dublin, Ireland

Local Arrangements
M. Marion, EPFL, Switzerland

M. Bach Cuadra, EPFL, Switzerland

G. Olmo, Politecnico di Torino, Italy

M. Gabbouj, Tampere UT, Finland

International Liaisons
B. Macq, UCL, Belgium

U.B. Desai,
Indian Inst. of Tech., India

D. Erdogmus,
Oregon H&S University, USA

J. Reichel, Spinetix S.A., Switzerland,

Copyright © 2007 Kenzan Technologies & Eusipco 2008