Calendrier des événements

Partager sur :

ESSEC working Group on risk

Generative Models for Dynamic Graph Exploration

L'Institut des actuaires  s'associe au Working Group on Risk de l'ESSEC-CREAR pour sa prochaine conférence sur le thème : "Generative Models for Dynamic Graph Exploration" 

par Fabrice Rossi Professeur à l'Université Paris-Dauphine CEREMADE & MILES.

Participer à cette conférence vous permettra d'ajouter 6 points à votre score de Perfectionnement Professionnel Continu (PPC).

 

Abstract : Graphs provide interesting models for representing interactions between entities. While a large part of the literature is dedicated to the simple situation where interactions are static and binary, many real world situations ask for more complex models which account for the temporal dimension of the interactions as well as for their other characteristics. In this joint work with M. Corneli, P. Latouche and C. Bouveyron, we propose extensions of the stochastic block model to interaction data with a temporal component and with a textual one.

A classical way of representing temporal interactions consist in constructing a time series of graphs, where each graph aggregates the interactions over a given time interval. While this approach can produce interesting results, it cannot adapt to more complex schemes where a single time scale cannot capture the full temporal dynamic of the interactions. In our approach, we use a continuous time model which enables us to limit temporal aggregation to very short time intervals, mostly motivated by computational consideration. We propose in addition a variant of our model that can model text messages attached to interactions via a dynamic topic model. 
I will present in this talk an introduction to generative models for graphs with a gradual introduction of elements needed to handle continuous time and information attached to interactions. The models will be illustrated on two real world examples: a validation study using the bike sharing system of London and a more complex analysis of the Enron email database.
Jeudi 20 février 2020
12h30 - 13h30 (GMT +2)
ESSEC
campus La Défense (CNIT) - Salle 202
  • Gratuit


Inscriptions closes
Lieu

ESSEC

campus La Défense (CNIT) - Salle 202

Aucun commentaire

Vous devez être connecté pour laisser un commentaire. Connectez-vous.

Jeudi 20 février 2020
12h30 - 13h30 (GMT +2)
ESSEC
campus La Défense (CNIT) - Salle 202
  • Gratuit


Inscriptions closes
  • Ajouter à mon agenda