\magnification=1200
\baselineskip=20pt
\nopagenumbers
\font\big=cmr12 scaled \magstep2

\centerline{\bf STANFORD UNIVERSITY}
\centerline{\bf DEPARTMENT OF STATISTICS}
\centerline{\bf STATISTICS SEMINAR}
\bigskip

\baselineskip=12pt
\centerline{4:15 p.m., Tuesday, February 27, 2001}
\centerline{Sequoia Hall Rm. 200}

\bigskip
\baselineskip=15pt
\centerline{\sl J. S. Marron}
\centerline{\sl University of North Carolina}
\bigskip
\centerline{\bf High Dimension - Low Sample Size data analysis}

\bigskip


    In medical imaging applications of functional data
analysis, individual "data points" are often quite 
expensive to obtain.  This results in a sample size
that is usually much smaller than the dimensionality.  
Data of this type motivate the study of very high 
dimensional probability distributions, where simple
calculations show that the usual intuition for 
conceptualizing such data is very misleading.  These 
same ideas are used to form the basis of a new 
conceptual model for HDLSS data.  This new model is 
used to motivate a new discrimination method, whose 
effectiveness is illustrated in the context of a real
data set of Corpus Callosum shapes.

\bye