BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
X-WR-CALNAME:Dissertation Defense: Wenlong Jiang
X-WR-TIMEZONE:Eastern Time (US & Canada)
BEGIN:VEVENT
DTSTAMP:20260720T094327Z
UID:tag:localist.com\,2008:EventInstance_49081496703852
DTSTART:20250313T160000Z
DTEND:20250313T180000Z
DESCRIPTION:The title for the dissertation is "Topics in Network Analysis a
 nd Multivariate Statistics".  An overarching objective in contemporary sta
 tistical network analysis is extracting salient information from datasets 
 consisting of multiple networks. While numerous methods have been develope
 d for analyzing network-valued data\, key challenges remain in network pro
 perty estimation\, integrating multiple networks for classification and co
 nstructing networks. In this thesis\, we develop novel methodologies to ad
 dress these challenges. First\, while considerable efforts have been devot
 ed to node and network clustering\, comparatively less attention has been 
 given to connectivity estimation and parsimonious embedding dimension sele
 ction. We propose a method to simultaneously estimate a latent connectivit
 y matrix and its embedding dimensionality (rank) after first pre-estimatin
 g the number of communities and node cluster memberships. The proposed met
 hod provides accurate and robust dimensionality estimates. When exact memb
 ership recovery is possible and dimensionality is much smaller than the nu
 mber of communities\, it outperforms averaging-based methods for estimatin
 g connectivity and dimensionality. Second\, we explore the integration of 
 functional and structural brain networks for classification\, showing that
  network embeddings obtained from the MultiVERSE algorithm can improve cla
 ssification accuracy over using a single network type. Finally\, motivated
  in part by the observation that correlation-based gene networks may not e
 xhibit the expected scale-free property\, while networks constructed after
  adjusting for latent factors are more likely to be scale-free\, we develo
 p a novel longitudinal factor analysis method. The proposed method accurat
 ely estimates latent factors and captures dynamic changes in both factors 
 and loadings over time\, particularly in response to treatment effects. Th
 e method uses an Empirical Bayes matrix factorization approach\, allowing 
 both factors and loadings to evolve over time.\n\nAdvisor and Committee Ch
 air: Dr. Chris McKennan
GEO:40.441386;-79.954582
LOCATION:Wesley W. Posvar Hall\, Statistics Seminar Room
SUMMARY:Dissertation Defense: Wenlong Jiang
URL;VALUE=URI:https://calendar.pitt.edu/event/dissertation-defense-wenlong-
 jiang
CATEGORIES:Defenses
END:VEVENT
END:VCALENDAR