SETIA

We are pleased to invite all interested academics,  students and professionals to attend the second version of SETIA (Seminario de Teoría de la Información y sus Aplicaciones).

The event will take place on Wednesday, November 30, 2011,
in room Auditorio Principal, Casa Central, Universidad Técnica Federico Santa María, Valparaíso, Chile.

Registration has no cost, but requires sending an email to
setia.2011.info(at)gmail.com
with the subject “SETIA Registration“. In the body of the email please include your name and institution.

We will thank those who download and post the SETIA 2011 announcement in their campuses or departments.
We hope to see you all here at SETIA2011!

The preliminary program of the event is the following:

10:00 – 12:00, Tutorial:
“Sacando el Máximo de un Puñado de Muestras: Una Introducción a Compressed Sensing”
Dr. Milan S. Derpich
Department of Electronic Engineering,
Universidad Técnica Federico Santa María, Chile

Abstract:
Existen  innumerables aplicaciones en las que se desea reconstruir una señal aparentemente compleja a partir de una cantidad relativamente pequeña de mediciones o muestras. En el mundo de la ingeniería, el escenario mejor conocido que garantiza el éxito en esta tarea es el teorema del muestreo de Shannon (-Whittaker-Kotel’nikov): si la transformada de Fourier de la señal de interés es cero más allá de una cierta frecuencia f, entonces se puede lograr reconstrucción perfecta tomando muestras regularmente a  frecuencia 2f.  De manera más general, en diversas situaciones  las señales de interés pueden representarse usando un número S de vectores de una base con N elementos, donde  S<<N (tal es el caso de, por ejemplo, imágenes naturales y bases  wavelet). Sin embargo, comúnmente no se sabe a priori cuáles son estos  vectores significativos, por lo que no resulta posible determinarlos directamente tomando sólo S mediciones. ¿Es necesario entonces tomar N muestras? La recientemente desarrollada teoría denominada  compressed sensing nos demuestra que, felizmente, la respuesta es no: con casi certeza, ¡basta con tomar un número de muestras proporcional a S log(N) para obtener reconstrucción perfecta!. Estos resultados se están aplicando exitosamente en áreas como el procesamiento de señales biomédicas,  geología y comunicaciones inalámbricas, entre otras.
En este tutorial, orientado a quienes no poseen conocimiento previo sobre el tema, se explicará la idea básica detrás de compressed sensing y se discutirán sus principales resultados, haciéndose referencia a los algoritmos más utilizados para su aplicación.

14:00-14:30, Talk:
“Self-modeling Robotics Using the Minimum Description Length Principle”
Dr. (c) Patricio Parada
Department of Electrical Engineering
Facultad de Ciencias Físicas y Matemáticas
Universidad de Chile

Abstract:
In this talk we will present a new application of the Minimum Description Length (MDL) principle for empowering a robot with self-modeling capabilities, a basic feature of a resilient robot. MDL is an statistical learning technique that was originally developed for the universal source coding problem. It establishes rules for computing the estimates of both the parameters and the structure of a model using sample values, avoiding the dangers of overfitting that other learning techniques have.
Our implementation uses a combination of mechanical measurements (speed, height, force applied by the servo motors) and preset exogenous stimuli to simulate the robot behavior under different conditions. The research working hypothesis is that the signals produced by the robot sensors are realizations of an independent stochastic process that depends of the actual morphology of the robot. If a leg is severed, for example, it will impact its underlying probability measure and will produce a new model. Our initial findings confirm that this assumption, conditioned to the class of measurements and the signal length under study, is sufficient for discriminating different models and that the results are more robust that those produced by other techniques used in this subfield of robotics.

14:30 – 15:00, Talk:
“On the Minimal Average Data-Rate that Guarantees a Given Closed-Loop Performance Level”
Dr. Eduardo I. Silva
Department of Electronic Engineering
Universidad Técnica Federico Santa María, Chile

Abstract:
This talks presents novel results related to  control system design subject to average data-rate constraints. By focusing on SISO LTI plants, and a class of source coding schemes, we establish lower and upper bounds on the minimal average data-rate needed to achieve a prescribed performance level. We also guarantee the existence of a specific source coding scheme, within the proposed class, that achieves the desired performance level at average data-rates below our upper bound. Our results are based upon a recently proposed framework to address control problems subject to average data-rate constraints.

15:00 – 15:20, Coffee Break

15:20 – 15:50, Talk:
“Rateless Coding for Temporally Correlated Optical Wireless Communication Channels”
Dr. Jaime A. Anguita
College of Engineering and Applied Sciences
Universidad de Los Andes, Chile

Abstract:
We present a demonstration of two error-correction coding schemes that can successfully operate on wireless optical communication channels subject to atmospheric turbulence. The codes (a punctured low-density parity-check code and Raptor code) operate by continuously adapting the information rate to accommodate the varying channel conditions. Because these coding schemes require the use of a feedback channel, we evaluate the bandwidth cost incurred. The evaluation of the codes is done over experimental optical signals recorded from a line-of-sight laser communication link. The temporal characteristics of the channels are analyzed and the performance of the codes are assessed under perfect and imperfect channel state information.

15:50 – 16:20, Talk:
“Analysis and Design of Wavelet-Packet Cepstral Coefficients for Automatic Speech Recognition”
Dr. Jorge Silva
Department of Electrical Engineering
Facultad de Ciencias Físicas y Matemáticas
Universidad de Chile

Abstract:
This work proposes using Wavelet-Packet Cepstral coefficients (WPPCs) as an alternative way to do filter-bank energy-based feature extraction (FE) for automatic speech recognition (ASR). The minimum probability of error signal representation (MPE-SR) criterion is adopted for selecting the filter structure that achieves an optimal tradeoff between feature complexity and discrimination. The problem is formulated as a complexity regularized optimization criterion, where the tree-indexed structure of the WPs is explored to find conditions for reducing this criterion to a type of minimum cost tree pruning, a method well understood in regression and classification trees (CART). In the experimental side, concrete filter-bank design considerations are stipulated to obtain most of the phone-discriminating information embedded in the speech signal, where the filter-bank frequency selectivity, and better discrimination in the lower frequency range [200Hz-1KHz] of the acoustic spectrum are important aspects to consider.

16:20 – 16:50, Talk:
“Maximum Expected Rates of Block-Fading Channels with Entropy-Constrained Channel State Feedback”
Dr. Milan S. Derpich
Department of Electronic Engineering,
Universidad Técnica Federico Santa María, Chile

Abstract:
We obtain the maximum average data rates achievable over block-fading channels when the receiver has perfect channel state information (CSI), and only an entropy-constrained quantized approximation of this CSI is available at the transmitter. We assume channel gains in consecutive blocks are independent and identically distributed and consider a short term power constraint. Our analysis is valid for a wide variety of channel fading statistics, including Rician and Nakagami-m fading. For this situation, the problem translates into designing an optimal entropy-constrained quantizer to convey approximated CSI to the transmitter and to de?ne a rate-adaptation policy for the latter so as to maximize average downlink data rate. A numerical procedure is presented which yields the thresholds and reconstruction points of the optimal quantizer, together with the associated maximum average downlink rates, by finding the roots of a small set of scalar functions of two scalar arguments. Utilizing this procedure, it is found that achieving the maximum downlink average capacity C requires, in some cases, time sharing between two regimes. It is found that for an uplink entropy constraint H < log_2(L), a quantizer with more than L cells provides only a small capacity increase, especially at high SNRs.

16:50 – 17:20, Talk:
“Incremental Refinement using a Gaussian Test Channel and MSE Distortion”
Dr. Jan Ostergaard
Aalborg University, Denmark

Abstract:
The additive rate-distortion function (ARDF) was developed in order to universally bound the rate loss in the Wyner-Ziv problem, and has since then been instrumental in e.g., bounding the rate loss in successive refinements, universal quantization, and other multi-terminal source coding settings. The ARDF is defined as the minimum mutual information over an additive test channel followed by estimation. In the limit of high resolution, the ADRF coincides with the true RDF for many sources and fidelity criterions. In the other extreme, i.e., the limit of low resolutions, the behavior of the ARDF has not previously been rigorously addressed.
In this work, we consider the special case of quadratic distortion and where the noise in the test channel is Gaussian distributed. We first establish a link to the I-MMSE relatation and use this to show that the slope of the ARDF near zero rate, converges to the slope of the Gaussian RDF near zero rate. We then consider the multiplicative rate loss of the ARDF, and show that for bursty sources it may be unbounded, contrary to the additive rate loss, which is upper bounded by 1/2 bit. We finally show that unconditional incremental refinement, i.e., where each refinement is encoded independently of the other refinements, is ARDF optimal in the limit of low resolution, independently of the source distribution. Our results also reveal under which conditions linear estimation is ARDF optimal in the low rate regime.