Annonce

Les commentaires sont clos.

Post-doc position: Distributed and Time-Adaptive Estimation of Communities in Graphs

31 Mars 2022


Catégorie : Post-doctorant


Conditions: 1 year of post-doctoral position opens, begining possible between June and September 2022, for a duration of 12 months.

Location: SISYPH team, Laboratoire de Physique de l’ENS Lyon 46 allee d’Italie, 69364 Lyon cedex 07

Advisors: Pierre Borgnat & Patrice Abry & Paulo Goncalves (from LIP, Inria) Contact: pierre.borgnat@ens-lyon.fr
web : http://perso.ens-lyon.fr/pierre.borgnat

 

Topic: Community detection in networks [1, 2, 3] is an important and well known topic which, in the past 20 years, as been tackled from many points of view in static graphs: statistical-based methods, physically inspired approaches, information-theoric views, compression-base schemes, random walk approaches, graphs signal processing insights [4]... The objective of community detection in networks is to find groups of nodes that are well connected together. In many com- plex systems, data are naturally represented as networks (or weighted graphs): social networks, sensor networks, Internet networks, neuronal networks, transportation networks, biological net- works... Also, accessing to a community structure enables to leverage it to process data (or signals, attributes) linked to the elements of the networks (it could be to nodes, edges or even higher-order structures such as multiplets or simplicial complexes), e.g. using Graph Signal Processing [5, 6, 7, 8] or Machine Learning techniques on graphs [9, 10, 11]. Despite the many advances and the fact that community detection is done now using these routine tools, there still have some practical obstacles when trying to apply these methods: i) the size of the graphs is possibly large and scalability of methods is often a major issue ; ii) data are often temporal, in that they are not accessed all at the same time, and they can evolve with respect to time. To face these issues, the objective of the post-doc be to explore distributed and adaptive methods for this question.

 

Background and Skills: The candidate must have skills in some of the following areas: Signal and Image Processing, Data science, Probability, Statistics, Modeling, Machine Learning, and/or Network Science. He/She will have to work in the frame of the group, and with external collaborators of the DARLING ANR project in Nice, Grenoble and Paris.

Application: Applicants must send by email a CV and a statement of interest to Pierre Borgnat. For further information, candidate can contact us with questions related to this position.