Español
EnglishConferencia: Extracción Supervisada de Características mediante técnicas de Análisis de Múltiples Variables basadas en Núcleo
A cargo del Dr. Jerónimo Arenas-García. Universidad Carlos III de Madrid.
Presentación
En esta charla tendrá lugar el jueves 26 de noviembre de 2009, a las 11:30 en la sala de grados de la Escuela Politécnica Superior de la Universidad de Alcalá:
Conferencia
Feature extraction (FE) is a key component for most automatic learning systems, specially in applications with data of very high dimensionality, or with redundant variables. In this lecture I will review several multivariate analysis (MVA) techniques that are useful for FE in both classification and regression systems: Principal Component Analysis (PCA), Partial Least Squares (PLS), Canonical Correlation Analysis (CCA) and Orthonormalized PLS (OPLS). Nonlinear extensions will be derived by means of kernel methods, paying special attention to a sparse version of Kernel OPLS, which can significantly reduce the computacional burden both in training and operational phases. The discriminative power of these MVA methods will be illustrated on several examples, showing the superiority of kernel over linear versions, and the convenience of using the target data for training the feature extraction projections.
Resumen del historial de Jerónimo Arenas-García:
Jerónimo Arenas-García received the Telecommunication Engineer degree from Universidad Politécnica de Madrid, Spain, in 2000, and the Ph. Degree in telecommunications Technologies from Universidad Carlos III de Madrid, Spain, in 2004. After a postdoctoral stay at the Technical University of Denmark, he returned to Universidad Carlos III de Madrid, where he is currently a Lecturer of digital signal and information processing His current research interests include statistical learning theory, particularly in adaptive algorithms and advanced feature extraction techniques, and their applications in remote sensing data and multimedia information retrieval. He has (co) authored more than 40 papers in these areas. Dr. Arenas-García is a current member of the PASCAL 2 NoE, and of the IEEE Machine Learning for Signal Processing Technical Committee, and serves on the Program Committee of the CIP and MLSP workshops.