banner

Site menu:

Principal >> Curso 2009-2010 >> Conferencias

Conferencias del Curso Académico 2009-2010

Democratization: A method for scaling up data mining algorithms

Día Hora Lugar Imparte
21/05/2010 10:00 Salón de Grados de la ETSIIT, UGR
Nicolás García‐Pedrajas, Universidad de Cordoba

In the last few years the problems faced by data mining algorithms have been growing in size from a few thousands of samples to hundreds of thousands or millions. This has rendered many algorithms unusable due to their scaling problems. Scaling up data mining algorithms has become an important issue, because for those algorithms "size does matter"...
This talk is devoted to the methods for scaling up instance selection algorithms.
-Firstly, we report the three general approaches used so far for scaling up data mining algorithms: designing fast algorithms, partitioning the datasets and using relational representations. We present some of the most common methods based on these three approaches.
-Secondly, we present a new paradigm, the democratization of algorithms, which is able to scale up learning algorithms without any modification. It is based on a divide-and-conquer strategy coupled with the principle of the ensembles of classifiers consisting of combining weak solutions to obtain a strong one. This paradigm has been successfully applied to instance selection and feature selection with notable results in terms of reduction of the execution time of the algorithms.
-Finally, we present the conclusions of the talk and the open research lines in the field of scaling up data mining algorithms.

Engineering Hybrid Intelligent Systems: Architectures, Perspectives and Challenges

Día Hora Lugar Imparte
19/02/2010 09:30 Salón de Actos de la ETSIIT
Ajith Abraham
Machine Intelligence Research Labs Washington USA

Esta conferencia ha sido financiada por la ayuda de movilidad de profesores visitantes MAS2009-00066-P del Ministerio de Innovación y Ciencia.

Anomaly and Change Detection in Time series Data

Día Hora Lugar Imparte
19/02/2010 11:30 Salón de Actos de la ETSIIT
Vipin Kumar

Computer Science and Engineering Department
University of Minnesota USA

Se discutirán el problema general de la detección de cambios y anomalías en series de datos temporales, varios enfoques y técnicas y por último algunas aplicaciones específicas en el dominio de las ciencias de la tierra/predicción de clima.
Esta conferencia ha sido financiada por la ayuda de movilidad de profesores visitantes MAS2009-00066-P del Ministerio de Innovación y Ciencia.

The Ecological Approach to Autonomous Robotics

Día Hora Lugar Imparte
18/02/2010 11:30 Salón de Actos de la ETSIIT
Alessandro Saffiotti
AASS Mobile Robotics Lab University of Örebro, Suecia

Desde un punto de vista clásico de la robótica autónoma, los robots y el ambiente han sido dos entidades diferentes. El ambiente es habitualmente asumido como no determinístico y sólo parcialmente observable, de modo que los robots sólo pueden interactuar con él mediante sus sensores sensibles al ruido y sus poco fiables actuadores. Este punto de vista a menudo se asimila a un juego de dos jugadores antagónicos, en la que el robot tiene que encontrar una estrategia para lograr su objetivo, a pesar de las medidas adoptadas por el ambiente. Últimamente está surgiendo un enfoque más ambiental en el campo de la robótica autónoma en el que robots y ambiente son considerados partes del mismo sistema con un interés común hacia un objetivo global o un estado de equilibrio. En el enfoque ambiental, los dispositivos robóticos están distribuidos en el ambiente mediante sensores, actuadores, electrodomésticos inteligentes, objetos etiquetados con RFID u otros robots. Estos dispositivos pueden comunicarse y colaborar unos con otros mediante la transmisión de información o la ejecución de acciones. Con este enfoque las funcionalidades más complejas no se consiguen construyendo un robot extremadamente competente pero aislado, sino con varios robots sencillos y embebidos en el ambiente. Este enfoque ambiental está actualmente en la base de muchos proyectos, incluyendo sistemas de redes de robots, robots ubicuos, ambientes artificiales, el proyecto U-RT y PEIS-Ecology. En la charla, se describirá esta visión ambiental y se discutirán algunos de los desafíos tecnológicos y científicos que implica este enfoque. Se proveerán algunos ejemplos concretos usando el entorno PEIS-Ecology, desarrollado por el grupo de investigación del orador.
Esta conferencia ha sido financiada por la ayuda de movilidad de profesores visitantes MAS2009-00066-P del Ministerio de Innovación y Ciencia.

Towards Networked Knowledge

Día Hora Lugar Imparte
18/02/2010 09:45 Salón de Actos de la ETSIIT
Stefan Decker

Digital Enterprise Research Institute
Galway, Irlanda

A pesar de la enorme cantidad de información disponible en la Web, aun nos faltan medios para interconectar y enlazar con sentido esa información para elevarla desde el nivel de la información al nivel del conocimiento. Adicionalmente, nuevas fuentes de información acerca del mundo físico están disponibles gracias a las emergentes tecnologías basadas en sensores. Esta información necesita integrarse con la información actual en la Web y en sistemas de información que requieren núcleos formados por semánticas ligeras. En mi charla discutiré el potencial de un espacio de conocimiento global y qué investigación y tecnologías son necesarias para habilitar la visión del "Networked knowledge" En mi charla voy a presentar diferentes flujos de trabajo y resultados del Digital Enterprise Research Institute en Galway en esta materia, desde Software Social Semántico sobre el Escritorio Semántico hasta redes de sensores.
Esta conferencia ha sido financiada por la ayuda de movilidad de profesores visitantes MAS2009-00066-P del Ministerio de Innovación y Ciencia.

Fundamentals of Computational Intelligence

Día Hora Lugar Imparte
26/11/2009 09:30 Salón de Actos de la ETSIIT
Professor Kaoru Hirota
Department of Computational Intelligence and Systems Science
Interdisciplinary Graduate School of Science and Engineering
Tokyo Institute of Technology, Yokohama, Japan

This is a basic lecture using PPT material about Computational Intelligence. The contents are: What is CI? Traveling Salesman Problem, Production System, Search, 2-valued logic, fuzzy logic, fuzzy inference, industrial applications.

Casual communication between humans and robots based on speech

Día Hora Lugar Imparte
26/11/2009 11:30 Salón de Actos de la ETSIIT
Professor Kaoru Hirota
Department of Computational Intelligence and Systems Science
Interdisciplinary Graduate School of Science and Engineering
Tokyo Institute of Technology, Yokohama, Japan

The Mascot Robot System is developed as a part of the “Development Project for a Common Basis of Next-Generation Robots” sponsored by NEDO (New Energy and industrial technology Development Organization). The Speech Recognition Module is also provided for this project by NEC Corporation and NEDO. The main purpose of the mascot robot system is to perform experiments demonstrating the functioning of the speech recognition module mounted on household robots. The Mascot Robot System's functioning is demonstrated in an ordinary living room, where casual communication between 5 robots and 4 human beings (1 host, 2 guests, and 1 walk-in) is conducted based on speech recognition and mentality expression of eye robots. The experimental results are shown by DVD files.

Measurement of Dielectrophoretic Force (Femto-Newton) Generated on a Small Particle and Potential Application to Identification, Separation and Handling of Blood Cells

Día Hora Lugar Imparte
25/11/2009 17:30 Salón de Actos de la ETSIIT
Professor Takeshi Yamakawa
Department of Brain Science and Engineering
Graduate School of Life Science and Systems Engineering
Kyushu Institute of Technology, Kitakyushu, Japan

Dielectrophoresis (DEP) is the motion of a matter caused by polarization effects in a non-uniform electric field. Recently, studies on DEP are promoted in various medical fields, including separation and characterization of biological cells. It is very important but not easy to measure this motive force, so-called dielectrophoretic force (DEP force). In general, the DEP force generated by the designed electrodes is analyzed by the computer simulation that employs the finite element analysis method. However, it does not always calculate the correct DEP force. Therefore, we propose the method of measuring the DEP force accurately based on the null method. The experiment is conducted with the dielectrophoretic device (DEP device) in which microfabricated electrodes were formed and which contains the polystyrene particles exhibiting the negative DEP in distilled water in a non-uniform electric field. The displacement of the particle reaches a steady state 15 min. after the change of applied voltage v, angle or frequency f. The equilibrium state of the particle in a non-uniform electric field can be reached at any place by adjusting both of angles and , where is the "Pitch" angle and the "Yaw" angle for the device. The proposed method can measure the incredibly minute DEP force ranging from 25 fN (femto-Newton) at to 298 fN at , the accuracy of which is determined by the static friction and rolling friction between particles and the inner floor of the DEP device.

Selected topics from 40 years of research on speech and speaker recognition

Día Hora Lugar Imparte
25/11/2009 11:30 Salón de Actos de la ETSIIT
Professor Sadaoki Furui
Department of Computer Science
Graduate School of Information Science and Engineering
Tokyo Institute of Technology, Tokyo, Japan

This talk summarizes my 40 years of research on speech and speaker recognition, focusing on selected topics that I have investigated at NTT Laboratories, Bell Laboratories and Tokyo Institute of Technology with my colleagues and students. These topics include: the importance of spectral dynamics in speech perception; speaker recognition methods using statistical features, cepstral features, and HMM/GMM; text-prompted speaker recognition; speech recognition using dynamic features; Japanese LVCSR; robust speech recognition; spontaneous speech corpus construction and analysis; spontaneous speech recognition; automatic speech summarization; and WFST-based decoder development and its applications.

Hot Issues in Evolutionary Multiobjective Optimization

Día Hora Lugar Imparte
25/11/2009 09:30 Salón de Actos de la ETSIIT
Professor Hisao Ishibuchi
Department of Computer Science and Intelligent Systems
Osaka Prefecture University, Osaka, Japan

Evolutionary multiobjective optimization (EMO) is one of the most active research areas in the field of evolutionary computation. A number of EMO algorithms have been proposed and applied to various problems in the literature. The main characteristic feature of EMO algorithms is that a large number of Pareto optimal or near Pareto optimal solutions can be obtained by their single run. The point in the design of EMO algorithms is to strike a balance between the convergence of solutions toward the Pareto front and their diversity along the Pareto front. Well-known and frequently-used EMO algorithms such as NSGA-II and SPEA2 share three common ideas: Pareto dominance, diversity maintenance and elitism. Pareto dominance is used as the primary criterion in the fitness evaluation phase together with an additional secondary criterion for diversity maintenance. Non-dominated solutions in the current population are usually handled as elite solutions. Pareto dominance-based EMO algorithms work very well on two-objective continuous optimization problems. They do not, however, work well on many-objective problems with more than four objectives. This is mainly due to the decrease in the selection pressure toward the Pareto front by the increase in the number of objectives. EMO algorithms do not always work well on multiobjective combinatorial optimization problems, either...

A Recent Topic on Control and Robotics. From Fuzzy Systems to Brain System

Día Hora Lugar Imparte
25/11/2009 15:30 Salón de Actos de la ETSIIT
Professor Kazuo Tanaka
Department of Mechanical Engineering and Intelligent Systems
University of Electro-Communications, Tokyo, Japan

This lecture gives a comprehensive treatment of a recent topic on control and robotics. The central subject of this lecture is the importance of originality and creativity in research in spite of dealing with a wide variety of ranges from fuzzy system to brain system. The first topic of this lecture is the recently developed fuzzy model-based control framework based on a linear matrix inequality approach. The second topic is development of a biologically-inspired flapping robot and cyclogyro-based robots with unique mechanisms. The last topic is a near-future application of a wearable brain-machine interface to wheelchair control. These topics are lectured with graduate level mathematics and mechanics in addition to with a number of experimental movies.

Development of Minimally Invasive and Definitive Surgery for Epilepsy

Día Hora Lugar Imparte
24/11/2009 09:30 Salón de Actos de la ETSIIT
Professor Takeshi Yamakawa Department of Brain Science and Engineering
Graduate School of Life Science and Systems Engineering
Kyushu Institute of Technology, Kitakyushu, Japan

Up to 5% of people in the world may have at least one seizure in their lives and about 1% are suffered from seizures hindering their daily life. The seizures produce loss of awareness in some cases. Although the seizures can be suppresses by adequate medication for about 80% of them, the residual 20% are not affected by the drug. The effective treatment for these patients of drug-resistive epilepsy is a surgical treatment to remove the "epileptogenic focus"...

Introduction to automatic speech and speaker recognition

Día Hora Lugar Imparte
24/11/2009 15:30 Salón de Actos de la ETSIIT
Professor Sadaoki Furui
Department of Computer Science
Graduate School of Information Science and Engineering
Tokyo Institute of Technology, Tokyo, Japan

This talk introduces history, applications and principles of automatic speech and speaker recognition technology using computers. Major techniques include statistical pattern recognition frameworks, Cepstrum and its dynamic features, HMMs (hidden Markov models), statistical language models, and various methods for increasing robustness against speaker and environmental variations. Major speech recognition applications are conversational systems for accessing information services and systems for transcribing ubiquitous spoken documents. Major speaker recognition applications are access control and speaker labeling for speech data management. Although speech and speaker recognition technology has made significant progress in the past 10-20 years with the help of computer technology, there still remain many unsolved problems.

Multi-Objective Optimization and Evolutionary Computation

Día Hora Lugar Imparte
24/11/2009 17:30 Salón de Actos de la ETSIIT
Professor Hisao Ishibuchi
Department of Computer Science and Intelligent Systems
Osaka Prefecture University, Osaka, Japan

Almost all real-world problems from decision making, planning and learning to design, control and data mining involve multiple objectives. Those objectives are often conflicting with each other. For example, weight minimization and strength maximization are typical conflicting objectives in various design problems. The main characteristic feature of multiobjective problems is that they usually have a large number of different optimal solutions. Among those solutions, we can not say which solution is the best. Thus we need additional information when a single solution is to be chosen as the final solution for a multiobjective problem. It is usually assumed in the context of multiobjective decision making that the choice of the final solution is performed based on the decision maker’s preference. There are three approaches to the use of the decision maker’s preference: a priori, interactive, and posteriori approaches. In the a priori approach, the decision maker’s preference is used to transform the original multiobjective problem into a single-objective one. On the other hand, the interactive approach tries to find the most preferred solution by using the decision maker’s preference to compare different solutions. In the posteriori approach, the choice of a single final solution is performed by the decision maker after a large number of candidate solutions are found...

Unique Challenges in Fuzzy Control, Flying Robotics and Brain-Machine Interface

Día Hora Lugar Imparte
24/11/2009 11:30 Salón de Actos de la ETSIIT
Professor Kazuo Tanaka
Department of Mechanical Engineering and Intelligent Systems
University of Electro-Communications, Tokyo, Japan

This lecture gives a brief overview of unique challenges in fuzzy control, flying robotics and brain-machine interface in our laboratory at The University of Electro-Communications (UEC) Tokyo, Japan. The topic ranges from fuzzy control based on nonlinear control approach and flying robotics based on aerodynamics paradigms to the so-called brain-machine interface utilizing electroencephalogram. This lecture emphasizes the importance of originality and creativity, i.e., the importance of unique challenge, in successful research.

Conferencia inaugural del curso 2009/2010: Softcomputing Industrial: Experiencias

Día Hora Lugar Imparte
18/11/2009 17:30 Salón de grados de la ETSIIT
Gregorio I.  Sainz Palmero   Director de la división de Tecnologías de la Información y las Comunicaciones de la Fundación CARTIF y Profesor titular de la Universidad de Valladolid en el Departamento de Ingeniería de Sistemas y Automática en la Escuela de Ingenierías Industriales.

Se abordará la experiencia acumulada a través de la realización de  proyectos industriales donde el softcomputing ha sido propuesto como base de la solución, y una alternativa a otros métodos tradicionales de la industria. Se repasarán algunos proyectos ya realizados en distintos ámbitos y las propuestas actuales donde la división TIC de la Fundación CARTIF está trabajando. Por otro lado, también se abordarán las dificultades, de diversa índole,  que existen en el proceso de transferencia del mundo  científico-universitario al mundo industrial y viceversa.

Máster: Soft Computing y Sistemas Inteligentes
Mapa Web
sugerencias
Dpto. Ciencias de la Computación e Inteligencia Artificial
E.T.S. Ingeniería Informática y de Telecomunicaciones
Universidad de Granada
 
¡CSS Válido!   Valid XHTML 1.1   AAA validation