BEGIN:VCALENDAR
PRODID:-//Columba Systems Ltd//NONSGML CPNG/SpringViewer/ICal Output/3.3-
 M3//EN
VERSION:2.0
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
DTSTAMP:20261106T144313Z
DTSTART:20261118T140000Z
DTEND:20261118T150000Z
SUMMARY:Statistics Seminar - Mu Niu (University of Glasgow)
UID:{http://www.columbasystems.com/customers/uom/gpp/eventid/}nhx-mw3o9qb
 v-z5b7ue
DESCRIPTION:Title: Learning Geometry from Data: An Intrinsic Diffusion Ap
 proach to Generative Modelling\n\nAbstract: Many modern datasets\, inclu
 ding images\, remote sensing measurements\, and biomedical data\, are hi
 gh-dimensional but often concentrate near lower-dimensional manifolds wi
 th complex non-Euclidean geometry. Ignoring this geometric structure can
  limit the effectiveness of statistical and machine learning methods\, p
 articularly in sparse-data settings. In this talk\, I will present a dat
 a-driven framework for learning latent geometric structure from data and
  incorporating it into stochastic processes defined on learned manifolds
 . Within this framework\, I will focus on diffusion-based generative mod
 elling\, where geometric information is used to guide the diffusion dyna
 mics and improve generation in sparse-data settings. In particular\, I w
 ill introduce the Intrinsic Hybrid Latent Diffusion Model (ILDM)\, which
  combines probabilistic dimensionality reduction with score-based diffus
 ion on learned manifolds. ILDM employs a hybrid diffusion process that a
 daptively switches between Riemannian and Euclidean dynamics according t
 o mapping uncertainty\, allowing the model to exploit geometric informat
 ion where it is reliable while retaining robustness elsewhere. Experimen
 ts on COIL-100\, MNIST\, and cardiac MRI datasets demonstrate substantia
 l improvements in generation quality compared with standard diffusion an
 d latent diffusion models. The talk highlights how latent geometry\, sto
 chastic processes\, and diffusion models can be integrated to develop ro
 bust generative methods for complex data.
STATUS:TENTATIVE
TRANSP:TRANSPARENT
CLASS:PUBLIC
LOCATION:Frank Adams 2\, Alan Turing Building\, Manchester
END:VEVENT
END:VCALENDAR