Nanophotonic and nanoplasmonic platforms for bioimaging and biosensing

When : Tuesday, October 25, 2016 - 15:00
Speaker : Dr Domenico de Ceglia
Affiliation : National Research Council, Charles M. Bowden Laboratory - US Army, USA
Where : Aula Magna "A. Lepschy"
Description :

Abstract: Nanostructures enable amplitude and phase manipulation of light at the subwavelength scale. The renewed and ever-increasing interest in nanoplasmonic and nanophotonic research is motivated by the rapid advancements in nanofabrication technologies and characterization tools, by the easier access to powerful and fast computational resources and by the progress in related research fields, e.g., metamaterials science. The existing and potential applications of nanostructures are innumerable and span a wide range of fields, including energy harvesting, sensing devices, electro-optical devices, biomolecule detection, drug delivery, and many more. We will discuss linear and nonlinear optical phenomena and we will give some examples of sensing and imaging devices based on nanostrutures.

Graph summarization with quality guarantees

When : Monday, September 26, 2016 - 16:30
Speaker : Dr. Matteo Riondato
Affiliation : Two Sigma Investments, New York (USA) and Brown University, Providence (USA)
Where : Sala Riunioni 318 DEI/G
Description :

Abstract: We study the problem of graph summarization. Given a large graph we aim at producing a concise lossy representation (a summary) that can be stored in main memory and used to approximately answer queries about the original graph much faster than by using the exact representation. To quantify the dissimilarity between the original graph and a summary, we adopt the reconstruction error and the cut-norm error, inspired by the Frieze-Kannan variant of Szemeredi’s regularity Lemma. By exposing a connection between graph summarization and geometric clustering problems (i.e., k-means and k-median), we develop the first polynomial-time approximation algorithms to compute the best possible summary of a certain size under both measures. Using the summary to answer queries is very efficient as the running time to compute the answer depends on the number of supernodes in the summary, rather than the number of nodes in the original graph. This is joint work with Francesco Bonchi and David Garcia-Soriano, published in the Data Mining and Knowledge Discovery journal.

A little knowledge is truly a dangerous thing

When : Tuesday, July 19, 2016 - 11:00
Speaker : Dott. Stefano Rini
Affiliation : National Chiao Tung University, Taiwan
Where : Aula Magna 'A. Lepschy'
Description :

Abstract: Interference pre-cancellation is a fundamental coding technique for multi-terminal networks and many classic information theoretic results rely on this strategy. Despite this, implementations of this coding scheme have yet to find widespread application in practical communication networks and one is bound to wonder as of why this is the case. In this talk we argue that interference pre-cancellation is effective only when very precise channel estimates are available at the users, a condition which is not easily attainable in many communication networks. The ability of a transmitter to pre-cancel the interference experienced at the receiver is drastically reduced when only partial side channel information is available. For this reason, we investigate the topic of robust interference cancellation and show that substantial coding advantages can be attained only when exploiting the structure of the interfering signal.

Optical Flow Switching: A Dynamic, Fast and Agile On-Demand Optical Network Architecture

When : Friday, July 15, 2016 - 11:00
Speaker : Vincent W.S. Chan
Affiliation : Massachusetts Institute of Technology, USA
Where : Aula Magna "A. Lepschy"
Description :

Abstract. Present-day networks are challenged by dramatic increases in data rate demand of emerging applications. With the advancement of high speed fiber communications with per wavelength transfer rate of 100Gbps or more and free space optical links for close-in data/computing center interconnects, unscheduled high speed data transfer of "elephant" (large volume) transactions are a reality. To make full use of the technology of optics and the very high data transfer rates, new network architectures from the physical to the transport layer are needed. This talk will address new network architectures, free of constraints of previous architectures and protocols, that will make "elephant" transactions much more efficient with much less network delays. Substantial performance gains can be expected for these new networks not just because the physical layer is faster but the upper layer protocols are also much improved in performance due to radical changes in architectures.

Grouping Games: Finding Clusters in Graphs, Digraphs and Hypergraphs

When : Monday, July 4, 2016 - 15:00
Speaker : Prof. Marcello Pelillo
Affiliation : Universita' Ca' Foscari Venezia
Where : Aula Magna "A. Lepschy"
Description :

Abstract: Clustering refers to the process of extracting maximally coherent groups from a set of objects using pairwise, or high-order, similarities. Traditional approaches to this problem are based on the idea of partitioning the input data into a predetermined number of classes, thereby obtaining the clusters as a by-product of the partitioning process. In this talk, I'll provide a brief review of recent work done in my group which offers a radically different view of the problem. In contrast to the classical approach, in fact, we attempt to provide a meaningful formalization of the very notion of a cluster and we show that game theory offers an attractive and unexplored perspective that serves well our purpose. To this end, we formulate the clustering problem in terms of a non-cooperative “clustering game” and show that a natural notion of a cluster turns out to be equivalent to a classical (evolutionary) game-theoretic equilibrium concept. We prove that the problem of finding the equilibria of our clustering game is equivalent to locally optimizing a polynomial function over the standard simplex, and we provide a discrete-time dynamics to perform this optimization, based on the Baum-Eagon inequality. The proposed grouping framework, which has already found applications in a variety of application fields, including computer vision, security and video surveillance, bioinformatics, etc., is general and can be applied to weighted graphs, digraphs and hypergraphs alike.

Galileo Satellite Navigation System: Current status and research opportunities

When : Tuesday, June 14, 2016 - 12:00
Speaker : Oscar Pozzobon
Affiliation : Qascom S.r.l., Italy
Where : Aula Magna "A. Lepschy"
Description :

Abstract: With the successful recent launch of VS015 mission, Galileo, The European Satellite Navigation System, can now count of a fleet of 13 operational satellites. After the initial delays and issues that affected the mission schedule, the Galileo system is now deployed with full ground segment support and full service capability will be declared soon. Two ground control centers, one in Fucino, Italy and one in hoberpraffenhfen, Germany, monitor and control both the mission data and space operations. Several sensor stations acquire, monitor and track the Galileo signals all over the globe. 5 uplink stations connect the ground segment to the space segment. Galileo will operate 4 different services with its signal, and specifically the open service (OS), The Commercial Service (CS), The Public Regulated Service (PRS) and Search and Rescue Service (SAR). The seminar will cover both the history in the system design, the challenges, the current status and how and when Europe will be able to use Galileo in an operational context. Furthermore, the seminar will discuss the innovation of such system with respect to GPS, the main differences, and the global political and technical challenges in the Global Navigation Satellite Systems scenario. As a conclusion, the research and industrial opportunities of Galileo will be discussed.

When is Big Data Sufficiently Big? When is it Too Big? Sample Complexity, Uniform Convergence, and Generalization Error

When : Monday, June 6, 2016 - 15:00
Speaker : Prof. Eli Upfal
Affiliation : Dept. of Computer Science, Brown University (RI), USA
Where : Aula Magna “A. Lepschy”
Description :

ABSTRACT: What sample size is needed to accurately estimate a parameter? For a given parameter, the answer is straightforward. But in most advance data analysis the parameter of interest is not pre-defined, it is the outcome of a search in some parameter space. The difficulty here is that the question implies a potentially large multi-comparison analysis, and classical statistics has no general effective method to evaluate the confidence level of such estimate from a sample. We present computationally efficient methods for evaluating the sample complexity (error bound for a given sample size) in specific domains, relevant to massive data mining applications. Our work will build on the classic concept of uniform convergence, computed through explicit Rademacher average bounds. Surprisingly, we demonstrate that our methods, that have provable statistical guarantees, are more computationally efficient than known heuristics that have no guarantees on the quality of their solutions.

The Integration of Computational Medical Modeling and 3D Rapid Prototyping with Pre-Surgical Planning

When : Wednesday, April 13, 2016 - 13:00
Speaker : Prof. Paolo Gargiulo
Affiliation : Reykjavik University, Islanda
Where : Aula Magna "A. Lepschy"
Description :

Abstract:  The use of rapid prototyping and 3D computational modeling to recapitulate pathological tissues and morphologies has been recently gaining much interest in the fields of biomedical engineering and regenerative medicine. However, despite increasing clinical interest and support, there are still few examples of how these services and methodologies may successfully integrate research institutions with clinical practice. Since 2007, our team at the University Hospital of Iceland, Landspitali, has been developing and enhancing our own 3D modeling and rapid prototyping services for the support of pre-surgical planning. More than 200 surgeries in different clinical sub-specialties have been planned thus - all with considerable success. For example, in maxilla facial surgery, dramatic reductions in operation time have been documented due to the ability for surgeons to physically emulate the navigation of complex craniofacial morphologies prior to surgery. Additionally, in orthopedic surgery, multiple operations on the same patient have been avoided due the pre-operative planning of combinatory surgical procedures. The most important notion, however, has been in regards to the overall life-saving utility of our 3D modeling processes - particularly as a key strategic tool in complex neurosurgical and cardiosurgical cases.Our modeling and rapid prototyping work has become fully integrated into the clinical processes at the Icelandic National Hospital: surgeons send a request with a patient's social security number to the engineering team, who may then access the hospital PACS system and retrieve the respective patient's CT or MRI data. From here, and typically within 24 hours, our surgical model is printed and delivered to the surgeons for pre-operative planning. Recently, we have started to combine multiple complex anatomical features and functions in our 3D printed models, and we are currently working on integrating tractography and using surgical navigation systems to accomplish an even more detailed surgical plan and simulation.Our future works aim to explore the utility of these techniques in assisting other clinical specialties and likewise in more complex surgical procedures, such as the design of titanium prosthetic devices and the use of laser sintering machines to produce an implantable part. We are also in the process of developing other pre-surgical tools that operate under similar computational methods, such as pre-surgical applications for the assessment of total hip arthroplasty patients using customized finite element modeling for patient-specific fracture risk assessment. In all, the proven utility of our 3D modeling and rapid prototyping methods can serve as a valuable example for clinics around the world, and we aim to continue researching innovative computational methods to improve the impact of these services.

Techniques to detect evoked potentials in EEG. Teaching and research activities at Programa de Engenharia Biomedical, Federal University of Rio de Janeiro.

When : Tuesday, April 5, 2016 - 16:30
Speaker : Prof. Antonio Mauricio Ferreira
Affiliation : Universidade Federal do Rio de Janeiro, Brasil
Where : Aula Magna 'A. Lepschy'
Description :

Abstract: It is well known that an external stimulation may change the spectrum of the electroencephalogram (EEG).  These changes may be either time- or phase-locked to the stimulation.  An example of this latter occurs in the Evoke Potential exam, where intermittent, periodic stimulation is used in order to elicit a cortical response.  Since the responses are small in comparison with the ongoing EEG, averaging techniques are often required in order to reveal them.  Thus the stimuli are repeated and the corresponding epochs are averaged in order to enhance the EEG component that is accounted for by the stimulation. Spectral changes in EEG that are not phase synchronized with the stimuli, but lead to time-locked changes in powerspectra, cannot be obtained through the averaging techniques used for estimating the evoked potential.  Those changes occur during the time the subject undergoes external stimulation or is asked to carry out a certain task.  They are commonly referred as Event-Related Synchronization and Desynchronization (ERS/ERD), depending on whether there is an increase or decrease in a given frequency band. In this presentation methodologies developed at UFRJ for quantifying both time and phase locked spectral changes in the EEG during sensory stimulation is are described. 

Development of Wide-Angle Liquid Crystal Fovea Lens

When : Thursday, March 31, 2016 - 11:00
Speaker : Prof. Sota Shimizu
Affiliation : Keio University, Japan
Where : Aula Magna “A. Lepschy”
Description :

Abstract:In this presentation, the wide-angle liquid crystal fovea Lens is introduced. This special lens can not only achieve both high resolution and wide angle simultaneously in a single image, but also change its attention part having high resolution in its field of view without any mechanical part.  In addition, it can achieve multiple attention areas in the single image on the same time (more than the human eye).