Post-doctoral position in Lille – Mapping of pollutant concentrations by deep learning

24/04/2022 – 25/04/2022 all-day

Offre en lien avec l’Action/le Réseau : – — –/– — –

Laboratoire/Entreprise : IMT Nord Europe
Durée : 18 months
Contact :
Date limite de publication : 2022-04-24

Contexte :
As part of the recovery plan aimed at preserving employment in Research and Development in France, IMT Nord Europe and TERA Group recruit to meet the challenges of a collaborative project whose objective is to develop a real size demonstrator of the potential for measuring air quality in a given geographical area. The demonstrator must combine optimized deployment of sensors and real-time data processing to visualize the temporal dynamics at very fine resolution (less than one minute) and at high spatial resolution (less than 50 meters) of concentrations in PM (Particulate Matter) on the area.

Sujet :
The post-doc position aims to produce a map representing the near real-time spatialization of pollution at the scale of an agglomeration. The main tasks are as follows:

– Define a methodology to establish a deployment strategy (definition of the spatial and temporal dynamics of the sensors) according to the objective, the scale and the quality level of the mapping. This work will help to optimize the number of sensors / measurement points and their spatial distribution. If necessary, reference point(s) will be integrated in the delimited geographical area (via one or more high-performance measurement stations).

– Develop reconstruction methods based on the state of the art (kriging with Gaussian processes) and more recent tools, such as deep learning using neural network architectures. It will be necessary to take into account the quality of the information (uncertainties depending on the sensors) and the integration of exogenous data (weather, topology, etc.) to improve the quality of the mapping.

Profil du candidat :
PhD in data science, signal / image processing, artificial intelligence, computer science, applied mathematics or atmospheric science (with experience in data processing) obtained in 2019, 2020, 2021 or 2022.

Formation et compétences requises :
– First experience with spatio-temporal interpolation methods (kriging) or/and machine learning methods, in particular deep learning,

– Use of data science tools: Python language and deep learning framework (such as PyTorch, TensorFlow or Keras).

Adresse d’emploi :
Cité scientifique,
Rue Guglielmo Marconi,
59650 Villeneuve-d’Ascq

Document attaché : 202204011425_2022_Annonce_Postdoc_Cartographie_Polluants_18_mois.pdf