Bayesian approaches for the improvement of parametric maps in Positron Emission Tomography

Data e Ora : Thursday, December 20, 2007 - 15:00
Relatore : Ing. Giampaolo Tomasi
Affiliazione : DEI
Luogo : Aula Magna `A. Lepschy`
Descrizione :

The generation of parametric maps -i.e. maps that associate to each pixel of the image its kinetic parameters- is a crucial issue in Positron Emission Tomography. Unfortunately, the quality of such maps is sometimes poor, due to the high noise of data at pixel level. In this talk, different types of Bayesian methods, which improve the quality and reliability of parametric maps, will be described and their potentialities showed both on simulated and real data sets.


Searching Nearest Neighbours Through Cluster Pruning

Data e Ora : Monday, December 17, 2007 - 15:00
Relatore : Prof. Alessandro Panconesi
Affiliazione : Universit� di Roma La Sapienza
Luogo : Aula Magna `A. Lepschy`
Descrizione :

Given a set of points S and a query point q, the k-nearest neighbour (k-NN) problem is to find the k points in S that are closest to q. This geometric problem is ubiquitous in (web) information retrieval and efficient algorithms are hard to come by, in spite of much work done in the area. In this talk we will introduce a very simple random clustering (RC) technique and show that: (1) RC can be analyzed mathematically and strong, useful properties can be proven about it; (2) with real data, RC outperforms well-known solutions for k-NN such as p-spheres and rank aggregation. Our experiments show that RC is superior to these well-known methods for a variety of applications, including text retrieval and image processing. Our theorems explain why.
Joint work with: Flavio Chierichetti, Prabakhar Raghavan, Mauro Sozio, Alessandro Tiberi, and Eli Upfal


Fuzzy Temporal Reasoning

Data e Ora : Thursday, December 6, 2007 - 15:00
Relatore : Ing. Marco Falda
Affiliazione : DEI
Luogo : Aula Magna `A. Lepschy`
Descrizione :

Many applications developed in very different areas such as Planning, Natural Language Processing, Hardware Design and Bioinformatics, need to manage critical aspects involving time. The models investigated in Artificial Intelligence literature for Temporal Reasoning are mainly based on the Constraint Satisfaction Problem (CSP) paradigm. Some of them deal with the representation of quantitative aspects of temporal data in terms of metric temporal statements about points. Other approaches, such as Interval Algebra, deal with a qualitative representation of temporal information in terms of qualitative relationships between intervals.
In everyday life these two aspects (either quantitative or qualitative) of temporal knowledge are not distinguished and real data are often affected by imprecision and uncertainty. Therefore, it is important to establish expressive frameworks for representing complex temporal relationships in a unified way.
In this talk the problem of representing and reasoning with fuzzy temporal knowledge in a general and flexible manner is introduced and a model for integrating qualitative and quantitative fuzzy temporal constraints is discussed.


Efficient Data Dissemination in wireless pervasive networks

Data e Ora : Thursday, November 22, 2007 - 15:00
Relatore : Ing. Elena Fasolo
Affiliazione : DEI
Luogo : Aula Magna `A. Lepschy`
Descrizione :

Data Dissemination consists on spreading a large amount of information to all nodes belonging to a network. Thus it can be applied for different purposes in a lot of practical scenarios. The peculiar characteristics of the system in use make the definition of efficient data dissemination schemes an interesting and challenging goal. In particular, developing efficient algorithms for wireless ad hoc networks is still an open issues due to the broadcast nature of the channel and to the need of managing all data transmissions in a distributed way. The former leads to a lot of problems related to the channel contention, collisions and interference. The latter requires to define algorithms which exploit only local information of the network and which are scalable and robust to the node mobility.
In this presentation we investigate how data dissemination schemes can be enhanced by the use of network coding. Network coding is a recent paradigm applied at network layer to increase the throughput. A lot of studies showed its effectiveness form a theoretical point of view. We, instead, investigate practical aspects related to the implementation of data dissemination schemes based on network coding in realistic environments affected by interference, collisions, fading and so on.


Multiple Description Coding for Video Applications

Data e Ora : Thursday, November 15, 2007 - 15:00
Relatore : Ing. Ottavio Campana
Affiliazione : DEI
Luogo : Aula Magna `A. Lepschy`
Descrizione :

Transmission of encoded video sequences over unreliable networks usually requires the adoption of protection techniques to guarantee good reconstruction quality at the receiver side. Multiple Description Coding (MDC) strategies add reliability to real-time video applications, where retransmission is not possible and packet losses afflict several frames degrading the overall quality.
In Multiple Description Coding, the transmitter splits the original signal into several correlated streams, called descriptions, which are sent to the receiver through distinct channels. Whenever all descriptions are correctly received, the signal is completely decoded at full coding quality, while in case of partial loss of some descriptions, lost information is estimated exploiting the correlation between descriptions and a coarse-quality signal is reconstructed.
In this talk, after a brief introduction to video and multiple description coding, applications of MDC to the H.264/AVC standard and Scalable Video will be presented.


Evolutionary algorithm techniques for network optimization problems

Data e Ora : Thursday, November 8, 2007 - 15:00
Relatore : Dr. Alessio Botta
Affiliazione : IMT Lucca Institute for Advanced Studies, Lucca, Italy
Luogo : Aula Magna `A. Lepschy`
Descrizione :

Evolutionary algorithms provide an off-the-shelf set of meta-heuristic techniques that can be used to solve complex, constrained and non-linear problems, including traditional optimization ones, such as models for scheduling and routing in wireless networks. In this colloquium, we will show technical details on how a basic evolutionary algorithm can be tailored to suit network optimization problems. Further, we will quickly review advanced evolutionary algorithms topics, such as co-evolution, constraint handling and multi-objective optimization, and provide some hints on how these techniques may be used to perform scalable, distributed and robust network problem solving.


Tree Based Combinatorial Optimization Problems in Telecommunication Networks

Data e Ora : Thursday, October 25, 2007 - 15:00
Relatore : Prof. Bezalel Gavish
Affiliazione : Southern Methodist University, Dallas, TX (USA)
Luogo : Aula Magna `A. Lepschy`
Descrizione :

Modern Computer Communication Networks are having a profound impact on individuals and society, the demand for higher capacity and faster networking services to the home are increasing significantly over time. This is known as the last mile problem, how to bring broadband communication services to the home at reasonable costs. Several technologies have been developed to provide such services to the end user: They include cable modems and wideband services through the cable TV infrastructure, wireless services through the cellular and wireless based systems and DSL/ADSL and its different variants that is based on extensions of the existing wire based phone system.
The different technologies and the analysis of their advantages and disadvantages will be presented. Most of the technologies rely on the existing local TV cable or phone network to provide services to the home. In order to deploy the technologies efficiently in the local service system efficient algorithms have to be developed to solve existing and new combinatorial optimization problems on tree based networks. We will present several of those problems and highlight results obtained to some of them.
Open problems and extensions will be discussed.


Intelligent Sensor Networks: the role of messaging middleware for end-to end business solutions

Data e Ora : Wednesday, October 17, 2007 - 14:30
Relatore : Dr. Pierre R. Chevillat
Affiliazione : IBM Research GmbH, Zurich Research Laboratory, Switzerland
Luogo : Aula Magna `A. Lepschy`
Descrizione :

This talk will give an overview of research activities in wireless sensor networking at the IBM Zurich Research Laboratory. We develop and prototype wireless network protocols, messaging middleware, and low-power sensor platforms to connect sensors & actuators with IBM enterprise computing. A particular focus is on publish/subscribe messaging protocols which offer unique advantages, e.g. for resilience and scalability, and which facilitate the deployment of applications across heterogeneous wireless sensor networks. Our goal is to enable `end-to-end` solutions, e.g. for asset management, logistics applications, or utility metering. In addition, we are working on location-sensing concepts which have application in retail environments (ムStore of the Futureメ). Finally, we participate in EU-funded Project e-SENSE exploring the use of sensor networks to capture ambient intelligence for B3G mobile communication systems.


Cross-layer design in wireless data networks: concepts and application examples

Data e Ora : Thursday, October 11, 2007 - 15:00
Relatore : Ing. Simone Merlin
Affiliazione : DEI
Luogo : Aula Magna `A. Lepschy`
Descrizione :

The increasing demand of multimedia contents such as video streaming, voice over IP services, gaming, and the massive peer-to-peer phenomena, arise the need for very high performance communications networks. These applications do not only require high data rates but also `quality of service` (reliability, delay, security...) need be provided. Moreover, the convergence among different technologies is required for providing more integrated and flawless services. In order to fulfill all these requirements, next generation networks has to be more efficient, flexible and reconfigurable.
The layered protocol architecture based on the ISO/OSI and the TCP/IP stacks, though of fundamental importance in allowing for an easy design and for understanding the basic network functionalities, has been shown to impose great limitations to the design of optimized solutions. The investigation of the relations among different layers and the study of mechanisms which break the classic layering is usually referred to as `cross-layer design`.
In this presentation, an introduction to the concept of cross-layer design is provided according to the literature, both from an architectural and analytical point of view. Two specific cases are then presented. The first one uses an analytical framework to define an algorithm for the joint congestion control, channel allocation and scheduling in multi-channel ad-hoc networks where each node is provided with multiple interfaces. In the second one, a simple mechanism spanning PHY, MAC and LL layers is investigated in the context of cellular networks, addressing the problem of trading fairness for physical layer efficiency.


Foundations of Digital Libraries and the Information Systems Research at UFMG

Data e Ora : Friday, October 5, 2007 - 11:00
Relatore : Prof. Marcos Andr� Gon�alves
Affiliazione : Federal University of Minas Gerais, Belo Horizonte, M.G., Brazil
Luogo : Sala Riunioni III piano DEI/G
Descrizione :

In this talk I will introduce the field of Digital Libraries and explain it through the perspective of the 5S framework of Streams, Structures, Spaces, Scenarios, and Societies. I will also discuss the application of the framework to the definition of a Quality Model for Digital Libraries. Relationships between 5S and the DELOS Reference Model for digital Libraries which is in development will be discussed. Finally, I will briefly comment on some other research efforts that have been conducted in Digital Libraries and Information Retrieval at Federal University of Minas Gerais, Brazil.


Applications of Infrared Photon Counting with Superconducting Single-Photon Detectors (SSPD)

Data e Ora : Wednesday, July 4, 2007 - 17:00
Relatore : Prof. A. Sergienko
Affiliazione : Dept. of Electrical & Computer Engineering and Dept. of Physics, Boston University, Boston, USA
Luogo : Aula Magna `A. Lepschy`
Descrizione :

Superconducting single-photon detectors (SSPD) allowed us to conduct photon-counting experiments with very broad continuous spectral coverage from 0.4 till 2.5 microns. We discuss several advantages this new technology brings in telecommunication area of spectrum. The major impact can also be seen in the area of biomedical applications where using very low level of infrared light enable us to avoid any disturbance of the object under investigation.


Bringing Metro and Access Network Together: Long-Reach PONs

Data e Ora : Thursday, May 31, 2007 - 15:00
Relatore : Dr. Giuseppe Talli
Affiliazione : Tyndall National Institute, Ireland
Luogo : Aula Magna `A. Lepschy`
Descrizione :

In the state-of-the-art, the access and metro portions of the network are provided by separate systems. As bandwidths grow, this traditional approach of separate access and metro networks will become prohibitively expensive:
- from a capex viewpoint due to the large number of network elements and interfaces to interconnect them;
- from an opex viewpoint due to network design complexity, large number of network elements, large footprint and high electrical power consumption.
A new generation of photonic communication system, which integrates access and metro into one system, has recently attracted the interest of the research community and network operators. These architectures offer the potential to reduce bandwidth transport costs by eliminating the costs of the electronic interface between the access and the outer core/metro back-haul network. In order to achieve this for European national geographies requires a reach of ~100 km from the customer to the ~100 major service nodes in a typical network. The ~100 km reach is necessary to enable full coverage and to allow the option of dual parenting for resilience.
The deployment of point-to-point fibre from each customer to the service node up to ~100 km away would be prohibitively expensive. These architectures, usually called long-reach passive optical networks (PONs) or Super-PONs, will therefore use multi-wavelength, high split PONs to make efficient use of fibre. While the first generations of PONs are now standardized and commercially available, the most advanced of these (GPON and GE-PON) typically offer 2.4 Gbit/s or 1 Gbit/s downstream and ~1 Gbit/s upstream, shared between 32 customers via passive optical splitters and a time-division multiple access (TDMA) protocol, over a reach of up to 20 km. These new architectures aim at a new generation of PON with features totally beyond the capability of today`s PONs:
- bandwidth per customer of up to 10 Gbit/s downstream and 10 Gbit/s upstream each 10 Gbit/s wavelength is shared by up to 512 customers
- significant use of DWDM to provide further fibre efficiency in the metro with up to 32 wavelengths each carrying 10 Gbit/s; the project will therefore take a hybrid WDM/TDMA approach
- all-optical reach of 100 km using optical amplifiers
- no use of optical-electrical-optical conversions at intermediate locations.
In this seminar we will review some of the technical challenges in the physical layer of long-reach PONs and the solutions proposed to overcome them. Our latest experimental and modelling results, obtained also within the IST project PIEMAN (Photonic Integrated Extended Metro and Access Network), will be presented.


Codifica video scalabile: wavelets, codifica distribuita e soluzioni basate su H.264/AVC

Data e Ora : Thursday, May 3, 2007 - 15:00
Relatore : Lorenzo Cappellari
Affiliazione : DEI
Luogo : Aula Magna `A. Lepschy`
Descrizione :

L`attuale esigenza di disporre di uno stesso contenuto video in diverse situazioni di utilizzo sta stimolando una fervente ricerca nel campo della codifica video scalabile. Il flusso di dati compressi in uscita da un codificatore scalabile e` tale per cui, scegliendone opportune porzioni, e` possibile ottenere rappresentazioni efficacemente compresse dello stesso contenuto video a risoluzioni spaziali, risoluzioni temporali e/o qualita` inferiori. Nel seminario verranno dapprima presentate le tecniche di codifica video scalabile attualmente in analisi da parte dei gruppi MPEG (ISO/IEC) e VCEG (ITU-T) con lo scopo di rilasciare una versione scalabile dello standard di codifica video H.264/AVC. Verranno poi discusse le soluzioni alternative proposte in questo ambito dal nostro gruppo di ricerca. In particolare, verra` presentata una soluzione recentemente sviluppata basata sulla codifica video distribuita nel dominio della trasformata wavelet


Robust Video Transmission using Distributed Source Coding

Data e Ora : Thursday, April 26, 2007 - 15:00
Relatore : Simone Milani
Affiliazione : DEI
Luogo : Aula Magna `A. Lepschy`
Descrizione :

The appearing of heterogeneous network scenarios, where mobile communications play a significant role, has introduced the need of dealing with a higher percentage of losses and errors with respect to the traditional wired communications. The increased amount of losses occurring during video transmissions makes the traditional video coding techniques inadequate and requires the design of robust schemes that provide a high Quality-of-Service (QoS) to the end user.
A recently proposed class of coders apply the principles of Distributed Source Coding (DSC) to video coding. The designed encoders enable a robust coding of visual information requiring a limited complexity.
However, these DSC-based video encoders have been unable to reach a compression efficiency comparable to that of motion-compensated predictive coding based video codecs, such as H.264/AVC. This motivates us to focus on the investigation of efficient models for video data in order to improve the compression gain.
Such a DSC-based, compression-centric encoder is an important block towards building a robust DSC-based video coding framework.


Rearrangement task realization by multiple mobile robots with efficient calculation of task constraints

Data e Ora : Thursday, April 19, 2007 - 15:00
Relatore : Norisuke Fujii
Affiliazione : University of Tokyo School of Engineering, Dept of Precision Eng.
Luogo : Aula Magna `A. Lepschy`
Descrizione :

We address a rearrangement task by multiple robot in this paper. A rearrangement task has constraints regarding the order of the start, grasping and finish time. Calculating these constraints has a high computational cost. We propose a rearrangement method that calculates constraints efficiently. In our approach, not all constraints are calculated, but some of them will be calculated step by step. The proposed method is tested in a simulated environment with up to 4 mobile robots. The methods are compared, and the results indicate that the proposed method is superior.


Sperimentazione in laboratorio del controllo e navigazione di un veicolo spaziale autonomo per operazioni di モdocking and servicingヤ

Data e Ora : Wednesday, March 21, 2007 - 15:00
Relatore : Prof. Marcello Romano
Affiliazione : Naval Postgraduate School, Monterey, CA (USA)
Luogo : Sala Riunioni 301 DEI/A
Descrizione :

A laboratory test bed is introduced, which enables the hardware-in-the-loop simulation of the autonomous approach and docking of a chaser spacecraft to a target spacecraft of similar mass. The test bed consists of a chaser spacecraft and a target spacecraft simulators floating via air pads on a flat floor. The prototype docking interface mechanism of Defense Advanced Research Projects Agencyメs (DARPAメs) Orbital Express mission is integrated on the spacecraft simulators. Relative navigation of the chaser spacecraft is obtained by fusing the measurements from a single-camera vision sensor and an inertial measurement unit, through Kalman filters. The target is collaborative in the sense that a pattern of three infrared Light Emitting Diodes is mounted on it as reference for the relative navigation. Eight cold-gas on-off thrusters are used for the translation of the chaser vehicle. They are commanded using a non-linear control algorithm based on Schmitt triggers. Furthermore, a reaction wheel is used for the vehicle rotation with a proportional derivative linear control. Experimental results are presented of both autonomous proximity maneuver and autonomous docking of the chaser simulator to the non-floating target. Further analytical and experimental studies are currently on-going on the navigation and control of a multi-spacecraft system for on-orbit assembly and reconfiguration.


The Hiring Problem and the Stochastic Performance of モLake Wobegonヤ Strategies

Data e Ora : Monday, March 12, 2007 - 15:00
Relatore : Prof. Eli Upfal
Affiliazione : Brown University, Providence, RI (USA)
Luogo : Aula Magna `A Lepschy`
Descrizione :

We study an iteratively decision process modeled as a hiring problem, in which a growing company continuously interviews and decides whether to hire applicants. This problem is similar in spirit but quite different in analysis from the well-studied secretary hiring problem. It captures aspects of decision making under uncertainty, and specifically issues that arise in systems where items are bought or sold at negotiated prices in an imperfect information market.
We analyze natural strategies of hiring above the current average, considering both the mean and the median averages; we call these Lake Wobegon strategies. (In Garrison Keillorメs legendary Lake Wobegon: `all the women are strong, all the men are good-looking, and all the children are above average`.) Like the hiring problem itself, our strategies are intuitive, simple to describe, and amenable to mathematically and economically significant modifications. We demonstrate several intriguing behaviors of the two strategies. Specifically, we show dramatic differences between hiring above the mean and above the median. We also show that both strategies are intrinsically connected to the lognormal distribution, leading to only very weak concentration results and the marked importance of the first few hires on the overall outcome.
Joint work with A. Z. Broder (Yahoo! Research), A. Kirsch (Harvard), Ravi Kumar (Yahoo! Research) and Michael Mitzenmacher (Harvard).


On Cross-Layer Design of MIMO Ad Hoc Networks

Data e Ora : Thursday, January 18, 2007 - 15:00
Relatore : Dr. P. Casari
Affiliazione : DEI
Luogo : Aula Magna `A. Lepschy`
Descrizione :

Ad hoc networks are formed of a collection of parties that communicate without need for an underlying infrastructure, and sharing some common medium, typically a radio channel within a specified bandwidth. Ad hoc networks are considered an important technology to improve communication flexibility, yet they suffer from a number of problems, e.g., the harshness of radio communications, and the increased difficulty of effectively access a shared medium.
Among the techniques to improve performance in ad hoc networks, the use of multiple antennas is receiving a lot of interest. At the price of increased signal processing complexity, multiple antennas bring into the picture some advantages, including but not limited to, directive long-range communications, spatial signal rejection, and Multiple-Input-Multiple-Output signaling, thus space-time coding, spatial multiplexing, interference detection and cancellation.
When joining such novel capabilities with existing protocols, the suboptimality of a simple change of the physical (PHY) level becomes apparent. In this talk, I will present a way to efficiently integrate PHY and Medium Access Control (MAC) in order to achieve better performance through cross-layer interactions. The obtained results are interesting and suggest to continue in this direction, extending the same methodology to multi-hop routing.


Biology-based Controller via Structural Transition of Nonlinear Oscillator Network

Data e Ora : Thursday, January 11, 2007 - 15:00
Relatore : Dr. T. Funato
Affiliazione : Tokyo Institute of Technology, Japan
Luogo : Aula Magna `A. Lepschy`
Descrizione :

In this talk, the construction of a structure-based controller is introduced. Natural controllers, represented by insect brain, show a fruitful adaptability with limited resources. Geometrical characteristics, particularly utilized for Passive dynamic walker and Small World structure, is possible origin of this functionality. In our method, multiple nonlinear oscillators are arranged as network elements, and synchronism is controlled by manipulating the connection structure. We first bring a coupled van der Pol oscillators into play, develop a controller based on eigenf-requencies of oscillators. Then we expand it to further general system, whose operation is independent from eigen-frequencies. Moreover, we apply this controller for a robot, which models cricket brain and changes the behaviour based on inner structure.