BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
X-WR-CALNAME:Robust Statistical Procedures for Clustering in High Dimension
 s
X-WR-TIMEZONE:Eastern Time (US & Canada)
BEGIN:VEVENT
DTSTAMP:20260516T060419Z
UID:tag:localist.com\,2008:EventInstance_31581544688493
DTSTART:20191018T170000Z
DTEND:20191018T180000Z
DESCRIPTION:Speaker: Dr. Anna Little\n\nAbstract: This talk addresses multi
 ple topics related to robust\nstatistical procedures for clustering in hig
 h dimensions\, including\npath-based spectral clustering (a new method)\, 
 classical\nmultidimensional scaling (an old method)\, and clustering in si
 gnal\nprocessing. Path-based spectral clustering is a novel approach which
 \ncombines a data driven metric with graph-based clustering. Using a data\
 ndriven metric allows for fast algorithms and strong theoretical\nguarante
 es when clusters concentrate around low-dimensional sets.\nAnother approac
 h to high-dimensional clustering is classical\nmultidimensional scaling (C
 MDS)\, a dimension reduction technique widely\npopular across disciplines 
 due to its simplicity and generality. CMDS\nfollowed by a simple clusterin
 g algorithm can exactly recover all\ncluster labels with high probability 
 when the signal to noise ratio is\nhigh enough. However\, scaling conditio
 ns become increasingly restrictive\nas the ambient dimension increases\, i
 llustrating the need for robust\nunbiasing procedures in high dimensions. 
  Clustering in signal\nprocessing is the final topic\; in this context eac
 h data point\ncorresponds to a corrupted signal. The classic multireferenc
 e alignment\nproblem is generalized to include random dilation in addition
  to random\ntranslation and additive noise\, and a wavelet based approach 
 is used to\ndefine an unbiased representation of the target signal(s) whic
 h is\nrobust to high frequency perturbations.
GEO:47.118149;-88.546013
LOCATION:Fisher Hall\, 327B
SUMMARY:Robust Statistical Procedures for Clustering in High Dimensions
URL;VALUE=URI:https://events.mtu.edu/event/robust_statistical_procedures_fo
 r_clustering_in_high_dimensions
CATEGORIES:Academics
CATEGORIES:Lectures/Seminars
END:VEVENT
END:VCALENDAR