Description : |
To model the organization of transcription regulatory elements along the
genome we have considered the use of multivariate point process models.
We have, in particular, used the multivariate, nonlinear Hawkes process.
This model class fits into a general framework of generalized linear point
process models, which we present.
We show a new theoretical result on the minimization of the penalized
minus-log-likelihood function over a Sobolev space, which in a special
case gives that the solution is in an explicit finite dimensional subspace
and in general gives an infinite dimensional descent algorithm.
For the nonlinear Hawkes process the parameters of interest are infinite
dimensional linear filter functions, and in the talk we discuss different
practical modeling strategies involving basis expansions and reproducing kernels. We illustrate the talk with applications to genome organization
based on data from ChIP-chip and ChIP-seq experiments.
|
Equipe organisatrice : |
Probabilités et Statistiques |
Jury (Thèse / HDR) : |
|
Orateur : |
Richard Hansen |
Titre (Thèse / HDR) : |
|
Université Orateur : |
Copenhague |
URL Orateur : |
|
Ressource : |
Laboratoire J.A.Dieudonné - Salle de conférence |
Date de début : |
11:30 - jeudi 15 avril 2010 |
Durée : |
1 heure(s) |
Date de fin : |
12:30 - jeudi 15 avril 2010 |
Type : |
Séminaire |
Réservation effectuée par : |
|
Dernière mise à jour : |
09:48 - mercredi 24 mars 2010 |