Se hace una cordial invitación a la charla: Optimal input design for nonlinear dynamical systems: a graph-theory approach, a dictar por el exalumno Sr. Patricio Valenzuela; para el día Jueves 17 de diciembre, a las 14.00 hrs., en el Auditorio "Guillermo Feick", B-221
Abstract: Optimal input design concerns the design of an input sequence to maximize the information retrieved from an experiment. The design of the input sequence is performed by optimizing a cost function related to the intended model application. Several approaches to input design have been proposed, with results mainly on linear models. Under the linear assumption of the model structure, the input design problem can be solved in the frequency domain, where the corresponding spectrum is optimized subject to power constraints. However, the optimization of the input spectrum using frequency domain techniques cannot include time-domain amplitude constraints, which could arise due to practical or safety reasons. In this talk, a new input design method for nonlinear models is introduced. The method considers the optimization of an input sequence as a realization of the stationary Markov process with finite memory. Assuming a finite set of possible values for the input, the feasible set of stationary processes can be described using graph theory, where de Bruijn graphs can be employed to describe the process. By using de Bruijn graphs, we can express any element in the set of stationary processes as a convex combination of the measures associated with the extreme points of the set. Therefore, by a suitable choice of the cost function, the resulting optimization problem is convex even for nonlinear models. In addition, since the input is restricted to a finite set of values, the proposed input design method can naturally handle amplitude constraints. Some applications of the proposed technique are presented to illustrate the discussion.
Patricio E. Valenzuela was born in Santiago, Chile in 1986. He obtained the electronics engineering title and M.S. degree in electronics engineering from the Universidad Técnica Federico Santa María, Valparaíso, Chile in 2011. In 2012 he joined the Department of Automatic Control at KTH Royal Institute of Technology, Stockholm, Sweden, where is currently pursuing a Ph.D. degree on Electronic Engineering. His research interests include system identification and control theory.
Se invita a la comunidad universitaria a la charla detallada a continuación, a ser dictada por el Prof. Rama Chellappa, de la Universidad de Maryland, EEUU, en el marco del Proyecto Basal AC3E:
-Título: "Deep Learning Networks for Computer Vision: Blessing or Curse?" -Fecha y hora: Viernes 11 de diciembre, 14:00 hrs -Lugar: Auditorio Guillermo Feick, B-221, Departamento de Electrónica -Expositor: Rama Chellappa, Chair, Department of Electrical and Computer Engineering University of Maryland https://www.umiacs.umd.edu/people/rama
ABSTRACT Over the last five years so, deep learning networks have shown that impressive results can be obtained for several computer vision problems such as face detection, object detection/recognition and face verification. In an ongoing effort on face verification, we have been able to improve performance using deep learning networks, which is both exhilarating and worrisome. While performance improvement using deep learning networks should be seen as a blessing (who can argue against performance), several problems remain to be addressed. For example, reliance on the availability of very large annotated data set may be a handicap. Being able to generalize networks across test data with different distributions (domain adaptation) as well as different problems is also important. Deriving bounds on the number of training sam! ples, gi ven the distance between training and test data distributions will be useful for planning data acquisition/annotation tasks. We will address some of these issues in the talk.
Prof. Rama Chellappa received the B.E. (Hons.) degree in Electronics and Communication Engineering from the University of Madras, India and the M.E. (with Distinction) degree from the Indian Institute of Science, Bangalore, India. He received the M.S.E.E. and Ph.D. Degrees in Electrical Engineering from Purdue University, West Lafayette, IN. During 1981-1991, he was a faculty member in the department of EE-Systems at University of Southern California (USC). Since 1991, he has been a Professor of Electrical and Computer Engineering (ECE) and an affiliate Professor of Computer Science at University of Maryland (UMD), College Park. He is also affiliated with the Center for Automation Research and the Institute for Advanced Computer Studies (Permanent Member) and is serving as the Chair of the ECE department. In 2005, he was named a Minta Martin Professor of Engineering. His current research interests span many areas in image processing, computer vision and pattern recognition. Prof. Chellappa is a recipient of an NSF Presidential Young Investigator Award and four IBM Faculty Development Awards. He received two paper awards and the K.S. Fu Prize from the International Association of Pattern Recognition (IAPR). He is a recipient of the Society, Technical Achievement and Meritorious Service Awards from the IEEE Signal Processing Society. He also received the Technical Achievement and Meritorious Service Awards from the IEEE Computer Society. He is a recipient of Excellence in teaching award from the School of Engineering at USC. At UMD, he received college and university level recognitions for research, teaching, innovation and mentoring undergraduate students. In 2010, he was recognized as an Outstanding ECE by Purdue University. Prof. Chellappa served as the Editor-in-Chief of IEEE Transactions on Pattern Analysis and Machine Intelligence and as the General and Technical Program Chair/Co-Chair for several IEEE international and national conferences and workshops. He is a Golden Core Member of the IEEE Computer Society, served as a Distinguished Lecturer of the IEEE Signal Processing Society and as the President of IEEE Biometrics Council. He is a Fellow of IEEE, IAPR, OSA, AAAS, ACM, AAAI and holds four patents.
El Centro Avanzado de Ingeniería Eléctrica y Electrónica (www.ac3e.cl) invita a la siguiente charla
“Formation Control of Vehicular Platoons" Dr. Andrés Peters Advanced Center for Electrical and Electronic Engineering UTFSM Fecha: Viernes 04 Diciembre 2015 - 11:30 hrs. Lugar: Auditorio Guillermo Feick - B-221
Abstract: Vehicular Platooning corresponds to a technique of optimizing the usage of limited resources such as road networks, fossil fuels, electricity and drivers time, among others, via the networked coordination and/or automation of vehicles. Although modern vehicles are equipped with certain levels of automation, the theory behind a multi-agent system such as a platoon of vehicles is not yet finished. Stability, String Stability and Performance are the most important theoretical concepts (from a mathematical point of view) to study in formation control motivated problems. They relate to the efficiency and safety of the systems and are commonly studied through a set of differential equations that model the interconnection of an arbitrary number of vehicles.
About the speaker: Andrés Peters was born in Coyhaique, Chile, in 1983. He obtained his Ingeniero Civil Electrónico and Magíster en Igeniería Electrónica degrees from the Universidad Técnica Federico Santa María of Valparaíso, Chile in 2007 and a Ph.D. in Applied Mathematics from the Hamilton Insitute at the National University of Ireland, Maynooth, Ireland in 2015. His research interests include performance bounds for optimal control of linear systems and string stability in formation control of vehicle platoons. He joined the Advanced Center for Electrical and Electronic Engineering, UTFSM, in October 2015.
Título: Procesamiento de imágenes para robots móviles Dr. Rodrigo Benenson, investigador Max Planck Institute de Informática, en Alemania.
Fecha: Martes 1 de Diciembre, 11:30 hrs. Lugar: Auditorio del Depto de Electrónica, B221
Resumen: La mayoría de los robots existentes están confinados en su área de operación. Para que un robot pueda moverse libremente en la cuidad necesita comprehender su ambiente para tomar las decisiones adecuadas. Las cámaras son sensores de bajo costo y al mismo tiempo una fuente rica de información. En esta charla voy a dar una perspectiva sobre los progresos en los últimos años en procesamiento de imágenes para robótica móvil, y los desafíos por venir. Voy también hablar brevemente sobre el vinculo entre investigación e innovación, y compartir un poco mi experiencia personal como sansano haciendo investigación en Europa.
Biografía: Rodrigo Benenson es un ex-alumno UTFSM de la generación elo99, titulado en 2004. En 2008 obtuvo un doctorado en robótica (Mines Paristech / INRIA, Francia), en el área de percepción para vehículos automáticos , y desde entonces ha seguido trabajando en el tema. Hizo un post-doc con Luc Van Gool en KULeuven, Bélgica (2009-2012), trabajando en procesamiento de imágenes; y actualmente es un investigador en el Max Planck Institute de Informática, en Alemania; trabajando con el profesor Bernt Schiele.
Rodrigo tiene múltiples publicaciones en conferencias internacionales (IROS, ICRA, ECCV, ICCV, CVPR, PAMI, etc.), acumulando 500+ citaciones.
El Departamento de Electrónica invita a la charla “Motion-based prediction with neuromorphic hardware”, a cargo del Dr. Laurent Perrinet, del Instituto de Neurociencias de la Timone, Francia, donde se desempeña como investigador de Neurociencia Computacional.