Organización  Novedades  Actividades  Fechas  Contribuciones  Programa  Inscripción  Estancia  Difusión  Restringido

Chuck Moore  
Corporate Fellow and the Technology Group CTO at Advanced Micro Devices

Parallel Computing in Next Generation System Architecture and Design

Abstract:
As Moore’s Law of transistor integration and the “Power Wall” continue to counteract one another, we are approaching a new era of system architecture and design. These systems will fundamentally embrace more power efficient data parallel computing structures, more autonomous concurrency, and more cost-effective scale-out approaches. They will take advantage of new technologies on the horizon that change fundamental bandwidth and latency ratios that we have build most traditional systems to tolerate. We will see that this era has significant implications on consumer devices, traditional personal computers, servers and even the highest performance supercomputers. In this talk, I will outline the driving forces behind these shifts, highlight the key issues, and comment on some of the possible solutions. As an end result of the talk, I hope to inspire the audience to move from the ongoing optimization of legacy systems, and into this exciting new era that will force much more system-level thinking and innovation. Indeed, we are headed into a very interesting decade of computer architecture!

Conferencia inaugural (plenaria)
Martes 7 Septiembre 2010, 19:15 h

Las próximas fronteras de la busqueda de informacion - The Next Frontiers of Search

Resumen
La búsqueda de información en la Web es ya una tecnología madura que hoy está acotada por el costo de la infraestructura y no por problemas a investigar. Por ende, los futuros desafíos de la investigación en tecnología de búsqueda para la Web tienen relación con la experiencia del usuario, la riqueza de los datos y aplicaciones novedesas. Esta conferencia analizará estas ideas.

Classical Web search is a mature technology that today is bound by technological investment rather than research issues. The next research challenges in Web search deal with the user, the richness of Web data and novel applications. In this talk we detail these ideas.

Conferencia Plenaria
Miércoles 8 Septiembre de 2010; 9:00 h

Camino del exaflop. Persiguiendo la mayor eficiencia energética y de espacio

Resumen
La modelización del clima y sus cambios, la actividad cerebral, la eficacia de los tratamientos médicos, la propagación de las ondas sísmicas, la dispersión de contaminantes en la atmósfera, la fusión nuclear, la descontaminación del agua, la prospección de petróleo, los análisis de riesgos en el sector financiero, el desarrollo de vehículos y aviones más fiables... son algunos de los ejemplos de las simulaciones que hoy se pueden realizar y que ilustran cómo la supercomputación de alto rendimiento se ha convertido en esencial para la investigación y la industria. Por esta razón Bull ha diseñado las soluciones bullx que se adaptan a todo tipo de aplicaciones y necesidades: desde un servidor departamental hasta el más innovador superordenador que incorpora la nueva generación de nodos SMP. Con sus soluciones Extreme Computing, Bull está trabajando en todos los frentes para reducir al mínimo el impacto medioambiental: componentes hardware de bajo consumo,  número reducido de componentes, innovadoras soluciones de refrigeración , procesadores eco-eficientes.

Conferencia Plenaria
Viernes 10 de Septiembre de 2010; 11:00
Lugar: Paraninfo, Edificio 3A

Detection of text plagiarism and Wikipedia vandalism

Abstract:
We will talk about the state of the art of computer-based methods that fight certain kinds of text misuse. It will be interesting to learn that, e.g. for plagiarism detection, there is no single solution approach: depending on the situation - analysis against the Web, analysis within a book, analysis among several papers - plagiarism detection means style analysis and authorship identification, but also cluster analysis, fingerprinting, or vertical search. In our talk we will also touch the emerging problem of Wikipedia vandalism, where we resort to crowdsourcing strategies using Amazon's Mechanical Turk. Not least the talk will provide some background about PAN, an evaluation campaign on plagiarism detection and vandalism detection, where the best research groups working in this field compare their detection technologies within an ambitious challenge.

S01-XXVI Congreso de la Sociedad Española para Procesamiento de lenguaje natural, SEPLN2010 (SEPLN)

Affective, persuasive and deceptive natural language processing

Absctract:
Dealing with creative language and in particular with affective,  persuasive and possibly humorous language has been considered outside the scope of computational linguistics. Nonetheless it is possible to exploit current NLP techniques starting some explorations about it. We briefly review some computational experience about these typical creative genres. In particular we will focus on research issues relevant to how affective meanings are expressed in natural language, and it will introduce techniques for affective content detection and generation. As far as persuasive language is concerned, after the introduction of a specifically tagged corpus (exploiting for example public political speeches), some techniques for persuasive lexicon extraction and for predicting persuasiveness of discourses are provided together with prospective scenarios of application. We conclude the talk showing some explorations in the automatic recognition of deceptive language.

S01-XXVI Congreso de la Sociedad Española para Procesamiento de lenguaje natural, SEPLN2010 (SEPLN)

Language Technology in action: computer-aided generation of multiple-choice tests

Abstract:
Natural Language Processing applications are far from perfect and computer-aided language processing with the user in control, is often regarded as an alternative and practical Language Technology approach to real-world implementations. This talk will present a computer-aided environment (tool) for generation of multiple-choice tests which can operate both in fully automatic and semi-automatic mode where the user is offered the option to post-edit the automatically generated test items. Given the time-consuming and labour-intensive task of developing multiple-choice tests manually, this environment has proven to be of help to test developers (e.g teachers, instructors).The presentation will introduce the underpinning original methodology for generation of multiple-choice tests from electronic documents. The tool based on this methodology reads the texts, identifies the important concepts and deems which sentences featuring such concepts can give rise to meaningful multiple-choice questions. Questions focusing on the important concepts are generated using transformational grammar rules and distractors are selected by identifying concepts which are semantically close to the correct answer. In addition to employing various NLP techniques including (but not limited to) term extraction and shallow parsing, the tool makes use of language resources such as corpora, Wikipedia and WordNet. The evaluation carried out shows that whereas the quality of the test items generated with the help of the tool (semi-automatic mode) is not compromised, much time is saved by using the tool to develop such items. The presentation will conclude with a recent study which investigateswhich strategy/metrics for selection of distractors based on semantic similarity, delivers best results.

S01-XXVI Congreso de la Sociedad Española para Procesamiento de lenguaje natural, SEPLN2010 (SEPLN)

Many-core platforms for embedded computing: 2012 and beyond

Abstract:
Programmability is a key requirement for fast  time-to-market and agile adaptation to rapidly evolving multimedia standards and customer expectations. Unfortunately, programmable architectures come with order-of-magnitude computational density and energy efficiency gaps with respect to custom-fit hardware. Is there a way to escape the flexibility vs. efficiency dualism?  Are nano-scale silicon technology and three-dimensional integration adding new facets to this "no free lunch" view? In this talk I will describe the architectural foundations and future evolutions of STMicroelectronics' Platform 2012 project and provide some insight on how we can hope to give positive answers to these fundamental questions.

S02-XXI Jornadas de Paralelismo, JP2010 (SARTECO)

Servidores de próxima generación para cálculo paralelo

Resumen:
HP presenta las nuevas generaciones en servidores orientados a cálculo paralelo, así como los proyectos de futuro en los laboratorios de HP. En esencia se trata de reducir al máximo el espacio y el consumo eléctrico pero aumentando el rendimiento y las facilidades de gestión. También se da cabida al cálculo híbrido CPU+GPU como solución a las necesidades de petaflops.

S02-XXI Jornadas de Paralelismo, JP2010 (SARTECO)

Display Holography: Historical Survey and Recent Developments

Abstract:
The world is currently witnessing an explosion of 3D digital data content, particularly in entertainment and geospatial applications; so why is the majority of this content being viewed on 2D screens? This talk will survey the history and fundamentals of digital holography displays, and survey a variety of different approaches that have been explored, with an emphasis on recent technologies and products, successful applications and future possibilities.

S03-XX Congreso Español de Informática Gráfica, CEIG2010
8 de septiembre a las 11

Stylized drawn animation: virtual rendering tools without the use of 3D models

Abstract:
Stylized drawn animation aims at creating animation in a painterly style and holds the promise of opening doors towards totally new animation techniques and styles such as aerographs, watercolors, pastels, gouaches, pencil effects... Digital paint simulation and animation methods allow us to pursue new pathways in this direction. This talk will highlight results obtained so far, covering both technical and productional aspects.

S03-XX Congreso Español de Informática Gráfica, CEIG2010
9 de septiembre a las 11

Distributed User Interfaces: How to Distribute User Interface Elements across Users, Platforms, and Environments

Abstract:
Distributed User Interfaces (DUIs) have become one vivid area of research and development in Human-Computer Interaction (HCI) where many dramatic changes occur in the way we can interact with interactive systems. DUIs attempt to surpass user interfaces that are manipulated only by a single end user, on the same computing platform, and in the same environment, with little or no variations among these axes. In contrast to such currently existing user interfaces, DUIs enable end users to distribute any user interface element, ranging from the largest one to the smallest one, across one or many of these dimensions at design and/ or run-time: across different users, across different computing platforms, and across different physical environments. In this way, end users could be engaged in distributed tasks that are regulated by distribution rules, many of them being currently used in the real world. This paper provides a conceptual framework that invites us to rethink traditional user interfaces in a distributed way based on the locus of distribution control: in the hands of the end user, under control of the system, or in mixed-initiative way. Any user interface submitted to distribution may also be subject to adaptation with respect to the user, the platform, and the environment.

S05-XI Congreso Internacional de Interacción Persona-Ordenador, Interacción 2010 (AIPO)

Interactive Systems not Interaction nor Interfaces

Abstract:
This talk proposes an extended view with respect to the paper from Michel Beaudouin-lafon at AVI 2004 entitled "Designing interaction, not interfaces". That paper argued that "the only way to significantly improve user interfaces is to shift the research focus from designing interfaces to designing interaction". Our presentation will focus on a broader view of this problem which is the design of Interactive Systems, interaction and interfaces being only the very small visible side of the iceberg. Addressing the problem requires the definition of methods, notations, processes and tools to go from early informal requirements to deployed and maintained operational interactive systems. The presentation will focus on critical interactive systems for which the cost of a failure is significantly higher than the development costs. The development of such interactive systems require the integration of knowledge from different computing science disciplines including reliability engineering, software engineering, resilience engineering and usability engineering. In that presentation we will present and integrated framework for the iterative design of operators' procedures and tasks, training material and processes and the interactive system itself. This integrated framework relies on an underlying formalism dedicated to the design, specification and validation of interactive systems and information about it can be found in Navarre et al. 2009.

S05-XI Congreso Internacional de Interacción Persona-Ordenador, Interacción 2010 (AIPO)

User Experience en el campo de eHealth

Resumen:
Cuando hablamos de eHealth nos referimos a la utilización de la información y tecnologías de comunicación para responder a las necesidades de los ciudadanos, pacientes, profesionales sanitarios, proveedores de atención médica y social, así como a responsables del sector de mercado. Es una etiqueta abreviada de la amplia gama de usos a los que las tecnologías de la información y comunicación se aplican en el ámbito sanitario y social. Abarca áreas de salud como la teleasistencia, telemedicina, telerehabilitación, telemática y telesalud. eHealth no es un grupo de productos, herramientas o aplicaciones, pero sí una serie de respuestas a un conjunto de requisitos en el contexto de la mejora y transformación de los servicios de salud. Esto es lo que se ha dado en llamar plataformas tecnológicas, que dan un marco de desarrollo de soluciones ad hoc para problemas concretos. Tanto el marco como cada solución estarán dirigidas por y para el usuario final o cliente. Desde un punto de vista empresarial, se trabaja en el desarrollo enfocado a soluciones para la monitorización y comunicación remota. El objetivo es crear productos basados en las necesidades y demandas de los usuarios, aplicando una metodología de evaluación de la experiencia del usuario de alto nivel en todas y cada una de las fases de desarrollo de un producto. Para ello, la experiencia de usuario no ha de limitarse a aspectos externos o de presentación, sino profundizar también en el diseño de sistemas que permitan a posteriori la localización de la plataforma a la demanda concreta del usuario, mercado o modelo de negocio.

S05-XI Congreso Internacional de Interacción Persona-Ordenador, Interacción 2010 (AIPO)

Property-based testing with QuickCheck

Abstract:
Since its first publication over ten years ago, QuickCheck has become the most widely used testing tool in the Haskell community, and been emulated in many other languages, including Ruby, Scala, and even Google's Go. It has spawned a commercial version in Erlang, and is at the heart of an EU project on property-based testing, ProTest. QuickCheck is appealing because it makes a simple and direct connection between program code and formal specifications, changing the focus of testing from inventing test cases to formulating properties--but it can be a challenge to apply in industrial practice! I will describe QuickCheck, including recent extensions and experiences, and present some new applications of the technology such as automatic generation of specifications.

S08-X Jornadas sobre Programación y Lenguajes, PROLE2010 (SISTEDES)

Metaheuristic hybridization with greedy randomized adaptive search procedures

Absctract:
GRASP, or greedy randomized adaptive search procedure, is a multi-start metaheuristic that repeatedly applies local search starting from solutions constructed by a randomized greedy algorithm. In this talk we consider ways to hybridize GRASP to create new and more effective metaheuristics. We consider several types of hybridizations: constructive procedures, enhanced local search, memory structures, and cost reformulations.

S11-VII Congreso Español sobre Metaheurísticas, Algoritmos Evolutivos y Bioinspirados, MAEB2010

Class Libraries and a Language Interface for Metaheuristics

Abstract:
Class libraries simultaneously provide a venue for describing taxonomies of heuristic search optimization methods and a code base upon which applications can be developed. In this talk we will review design issues and describe a few of the extant libraries. In contrast to exact methods, where commercial and open-source codes are heavily used, metaheuristics are often implemented without use of a library.  One reason for this seems to be a lack of an interfaces to algebraic modeling languages. We will describe connections between class libraries and modeling languages and examine the benefits and limitations to such partnerships.

S11-VII Congreso Español sobre Metaheurísticas, Algoritmos Evolutivos y Bioinspirados, MAEB2010

Clasificación multietiqueta

Resumen:
La clasificación multietiqueta consiste en asignar cada caso simultáneamente a varias clases, las cuales están relacionadas entre sí.  Hoy en día es uno de los problemas mas interesantes en minería de datos, con aplicaciones en dominios tales como la categorización de escenas, la categorización de texto o la bioinformática. En la charla se presentarán diversas formas en las que este problema ha sido abordado en la literatura, prestando especial atención a métodos basados en modelos gráficos probabilísticos.

S13-V Simposio de Teoría y Aplicaciones de Minería de Datos, TAMIDA2010
Jueves 9 de septiembre de 11:00 a 12:30

Cloud computing y su impacto en la informática

Abstract:
 
Cloud computing is in principle less a radical technical innovation than a change in the business model for IT infrastructures. However, this change in the business model is likely to cause deep and fundamental changes in the way we approach the design of software and data processing systems at all scales. In this talk I will overview the main motivation behind cloud computing (cost reduction and consolidation of IT infrastructures) through a number of examples from industry, the technical aspects that are driving cloud computing, and then proceed to explore how these developments will influence the design, deployment, and operation of IT infrastructures -including some thoughts on the impact on university education for computer scientists.

S04-XV Jornadas de Ingeniería del Software y Bases de Datos, JISBD2010 (SISTEDES)