Workshops

There is an ever-expanding range of applications areas where analysis of very high-dimensional datasets lies at the heart of the matter. Relatively little is known about the structure of naturally-occurring high-dimensional data, and there are enormous opportunities for high-impact discovery. Recently, a number of researchers have pointed out the relevance of multiscale approaches to understanding high-dimensional data. For example, some data manifolds, such as image manifolds, are inherently multiscale, and some methods, such as density estimation in high dimensions, can profit greatly by multiscale representations.

In this workshop we will explore the interactions between multiscale thinking and high-dimensional data, and we hope to bring together experts from classical machine learning, spectral graph theory, and visualization and sonification of high dimensional data, etc.

Topics to be covered include:

• Structure of high-dimensional data manifolds
• Multiscale structures in high-dimensional data
• Methods for dimensionality reduction
• Relations between machine learning, statistical and math analysis viewpoints.
• Visualization and sonification
• Multiscale Computational Algorithms

Organizing Committee

Ronald Coifman (Yale University)
David Donoho (Stanford University)
Mark Hansen (UCLA)
Peter Jones (Yale University)
Naoki Saito, Chair (University of California at Davis)