BEGIN:VCALENDAR
VERSION:2.0
CALSCALE:GREGORIAN
PRODID:iCalendar-Ruby
BEGIN:VEVENT
CATEGORIES:Colloquium
DESCRIPTION:Spectral Clustering in the Gaussian Mixture Block Model\n\nThe 
 Gaussian Mixture Block Model is a generative model for networks with commun
 ity structure\, designed to better capture structures observed in real-worl
 d networks. In this model\, each vertex is associated with a latent feature
  vector\, which is sampled from a mixture of Gaussians. These Gaussian comp
 onents correspond to distinct communities within the network. Between each 
 pair of vertices\, an edge is added if and only if their feature vectors ar
 e sufficiently similar.\n\nIn this talk\, I will present an efficient spect
 ral algorithm for clustering (inferring community labels) and embedding (es
 timating latent vectors). My focus will be on the high-dimensional regime\,
  where the latent feature space is high-dimensional – a setting that is bot
 h relevant to modern networks and mathematically challenging. For embedding
 \, I will demonstrate that\, provided the graph is not too sparse and the d
 imensionality is not excessively high\, the spectral algorithm delivers acc
 urate estimates of the latent vectors. Furthermore\, for clustering\, when 
 the separation between communities is sufficiently large\, the spectral alg
 orithm enables the recovery of the communities. I will also discuss corresp
 onding impossibility results\, highlighting conditions under which these ta
 sks become infeasible\, thereby rendering our results sharp up to logarithm
 ic factors.\n \n\nBio: Shuangping Li is a Stein Fellow in statistics at Sta
 nford University. She earned her Ph.D. in applied and computational mathema
 tics from Princeton University under the guidance of Professors Allan Sly a
 nd Emmanuel Abbe. Her research lies at the intersection of probability theo
 ry\, theory of algorithms and complexity\, high dimensional statistics\, an
 d theoretical machine learning.
DTEND:20250306T221500Z
DTSTAMP:20260416T112747Z
DTSTART:20250306T211500Z
GEO:42.443451;-76.481506
LOCATION:Frank H. T. Rhodes Hall\, 253
SEQUENCE:0
SUMMARY:ORIE Colloquium: Shuangping Li (Stanford)
UID:tag:localist.com\,2008:EventInstance_48985109299029
URL:https://events.cornell.edu/event/orie-colloquium-shuangping-li-stanford
END:VEVENT
END:VCALENDAR