The Weizmann Institute of Science
Faculty of Mathematics and Computer Science
Vision and Robotics Seminar
Boaz Nadler
will speak on
Learning in High Dimensions, Sparsity and Treelets
Abstract:
Scientific problems in many fields involve regression, classification or
clustering of datasets in very high dimensions ( $>$ 1000), sometimes with a
relatively small number of samples ($<$100). In many applications, however,
such as in spectroscopy, gene expression and document clustering, the original
data has a much lower intrinsic dimension. An interesting question is then what
is the performance of learning algorithms in this setting and how can one take
advantage of the intrinsic low dimensionality of the data.
In this talk I'll discuss three interconnected themes:
1) Problems in learning in high dimensional spaces,
2) Different notions of sparsity to overcome them,
3) Treelets - a new contruction of an adaptive multiscale basis for unordered
data.
Part of this talk is joint work with Ronald Coifman, and with Ann Lee and Larry
Wasserman.
The lecture will take place in the
Lecture Hall, Room 1, Ziskind Building
on Thursday, May 31, 2007
12:00 - 13:00