El 19 de septiembre de 2013 a las 15:30 h. en el Salón de Grados de la Escuela Politécnica Superior de la Universidad Autónoma de Madrid tiene lugar el seminario «Intrinsic dimensionality reduction with application to chemical simulation models», impartido por Neta Rabin, de Department of Exact Science at Afeka Academice College of Engineering, Tel-Aviv, Israel.
Simulation data from physical and chemical processes is typically high-dimensional. The dynamics of such systems can often be well described in fewer dimensions. In this talk we present a method for nonlinear dimensionality reduction that embeds a high-dimensional data set in a low-dimensional intrinsic space. The obtained reduced variables are constructed through the eigendecomposition of a Laplace operator. The Laplace operator is built by using a Riemannian metric, which captures the intrinsic geometry of the data. We apply the method to two simulated data sets resulting from different partial observations of the same underlying dynamic process and use the intrinsic variables to map the data into a common low-dimensional space. Furthermore, we demonstrate how to extend the partially observed high-dimensional data in the ambient space from the constructed reduced coordinates.
Neta Rabin. Received her B.Sc degree in Mathematics and Computer Science, and her M.Sc and Ph.D degrees in Computer Science from TelAviv University, Tel-Aviv, in 2001, 2004, and 2010, respectively. From 2010 to 2012, she was a Gibbs Assistant Professor in the Applied Mathematics Department at Yale University. Since October 2012 she is a faculty member in the Department of Exact Science at Afeka Academice College of Engineering, Tel-Aviv, Israel. Her research interests include high dimensional data analysis, manifold learning and signal processing.