• Home • Mission & Research • History • People • Scientific Staff • Supporting Staff • Scientific Advisory Board • Board of Directors • Event Calendar • Publications • Affiliation with the Helen Wills Neuroscience Institute at UC Berkeley • Employment Opportunities • Location & Contact Information ................................... Join Our Mailing List		HIGH-DIMENSIONAL DATA DAYS Friday and Saturday, February 21-22, 2003 Neurons deal with 10^3 to 10^5 inputs, most likely representing the relevant probability distributions in the brain’s sensory-motor world with their combined activity. Similar high-dimensional data arise, for example, with traffic patterns on the web, or in image databases. The ability to accurately learn a low-dimensional parameterization from high-dimensional data, and possibly to patch together a ‘manifold’, could revolutionize our understanding of the representational apparatus of the brain, and, for example, speed up web search engines through the use of a cached representation. Methodologies to address these high-dimensional problems are being developed in many different disciplines from psychology to computer science. This one day workshop features talks on foundational principles, representational techniques and discussion of future directions and problems, particularly the cross-disciplinary fertilization of neural representation theories by advanced machine learning techniques involving probabilistic learning, group theory and differential geometry.   Friday, February 21, 2003 9:00   Breakfast 9:30   Introduction Jeff Hawkins, Carrie Grimes & Tony Bell 10:00   "Learning High Dimensional Correspondences from Low Dimensional Manifolds" Lawrence Saul University of Pennsylvania 10:45   "Multi-layer DCA, Data Manifolds, Statistical Physics and Visual Processing in the Brain" Tony Bell Redwood Neuroscience Institute 11:30   "ICA Using Spacings Estimates of Entropy" Erik Miller University of California at Berkeley 12-12:30   Discussion 12:30-2:00   Lunch 2:00   "Geometric Harmonics: Can One Hear the Shape of a Dataset?" David Donoho Stanford University 2:35   "Color Correction for Digital Cameras" Brian Wandell & Jeffrey DiCarlo Stanford University 3:10   "Rejection Based Classifier for Target Detection In Images" Michael Elad Stanford University 3:45   Break 4:15   "Exploratory Segmentation and Fitting of a 3D Surface" John M. Agosta Intel Corporation 4:50   "Learning Object-Independent Modes of Variation with Feature Flow Fields" Erik Miller University of California at Berkeley   Saturday, February 22, 2003 9:00   Breakfast 9:30   Introduction Jeff Hawkins, Carrie Grimes & Tony Bell 10:00   "Function Estimation on Riemannian Manifolds" Partha Niyogi University of Chicago 10:45   "ISOMAP & LLE: Analysis of Recent Developments in Dimensional Reduction" Carrie Grimes Stanford University 11:30   "Topological Methods in Dimensionality Reduction" Vin de Silva Stanford University, Mathematics Department
			© 2002 Redwood Institute. All rights reserved.