BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.2.3.1//EN
TZID:Europe/Paris
X-WR-TIMEZONE:Europe/Paris
BEGIN:VEVENT
UID:8428@i2m.univ-amu.fr
DTSTART;TZID=Europe/Paris:20140321T140000
DTEND;TZID=Europe/Paris:20140321T150000
DTSTAMP:20241120T210410Z
URL:https://www.i2m.univ-amu.fr/evenements/g-rabusseau-lif-learning-negati
 ve-mixture-models-by-tensor-decomposition/
SUMMARY: (...): G. Rabusseau (LIF): Learning Negative Mixture Models by Ten
 sor Decomposition
DESCRIPTION:: Learning Negative Mixture Models by Tensor Decomposition by G
 uillaume Rabusseau (LIF)\n\nAbstract:\nIn this talk\\\,  we consider the p
 roblem of estimating the parameters of negative mixture models\\\, i.e. mi
 xture models that possibly involve negative weights. We show that every ra
 tional probability distributions on strings\\\, a representation which occ
 urs naturally in spectral learning\\\, can be computed by a negative mixtu
 re of at most two  probabilistic automata (or HMMs).  We present a method 
 to estimate the parameters of negative mixture models having a specific te
 nsor structure in their low order observable moments. Building upon a rece
 nt paper on tensor decompositions for learning latent variable models\\\, 
 we extend this work to the broader setting of tensors having a symmetric d
 ecomposition with positive and negative weights. This extension leads to a
  generalisation of the tensor power method for complex valued tensors\\\, 
 for which we establish theoretical convergence guarantees. Finally\\\, we 
 show how our approach applies to negative Gaussian mixture models.
END:VEVENT
BEGIN:VTIMEZONE
TZID:Europe/Paris
X-LIC-LOCATION:Europe/Paris
BEGIN:STANDARD
DTSTART:20131027T020000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
END:STANDARD
END:VTIMEZONE
END:VCALENDAR