Offre en lien avec l’Action/le Réseau : MACLEAN/– — –
Laboratoire/Entreprise : UMR TETIS / INRAE
Durée : CDI
Contact : dino.ienco@inrae.fr
Date limite de publication : 2022-03-08
Contexte :
Part of the research conducted within the TETIS JRC (Joint Research Unit) concerns the design, development and use of AI techniques especially tailored for the analysis of remote sensing data with the aim to meet the agro-environmental challenges raised up by the different institutions to which it belongs to (CIRAD, INRAE, AgroParisTech and CNRS). To strengthen these research activities and maintain an original scientific position at both national and international level, it is important to consolidate the role of the TETIS JRC in this strategic and innovative area.
You will be associated with the ATTOS team (Acquisition, Remote Sensing, Processing and Spatial Observations). The aim of the team is to shed light on the understanding of systems and territorial decision-making in a context of agro-ecological transition, demographics, and sustainable development through the exploitation of remote sensing and Earth observation data. You will reinforce the team’s axis associated with the design, development and use of AI techniques for the analysis of remote sensing data, to support different applications/themes at the core of the team and the unit (i.e. land use, natural resource mapping, biodiversity characterization, extraction of biophysical variables, monitoring of forest environments, study of artificialization, etc.) in tight collaboration with fellow researchers and engineers coming from a multidisciplinary context.
Your work will contribute to the positioning of the unit in the emerging research field at the crossroads between AI and remote sensing. This fast growing research field shows a high potential of scientific production, with the possibility of major societal advances concerning both the agriculture and the environment application domains at the core of the INRAE institute.
Sujet :
Your scientific activity will be oriented towards emerging paradigms in the machine learning and computer vision fields such as: (i) Spatial and/or temporal transfer learning, domain adaptation of AI models to cope with environmental diversity in acquisition conditions (ecosystems, climate, etc.) and temporal variability; (ii) Multi-source and multi-modal remote sensing data fusion (multi-modal and cross-modal learning) taking into account the variability in terms of availability of the input sources (sources unavailable for inference, syntheses, etc.) (iii) Limited availability/quality/reliability of reference data (weakly-supervised learning) and (iv) The need for interpretability of the developed machine learning models (explainable AI) and their combination with underlying physical models (physical-based machine learning) for the analysis of earth observation data.
https://jobs.inrae.fr/en/open-competitions/open-competions-research-scientists-job-profiles-crcn/cr-2022-mathnum-3
Profil du candidat :
You have a PhD, ideally in computer science or artificial intelligence.
You have competences in the foundations of machine learning methodologies (supervised, semi-supervised, unsupervised learning), knowledge and/or experience in modern deep learning paradigms (domain adaptation, transfer learning, weakly supervised learning and knowledge distillation), and
the development and exploitation of the latter in the context of signal processing and image analysis (computer vision).
Already proven experiences in the analysis of satellite and earth observation data with applications in the field of agriculture or environment would be appreciated.
Fluency in English is desirable, as well as long-term international experience: laureates who have not yet had such experience will be strongly encouraged to spend a period of time abroad, co-constructed with the host team, within 3 years after the internship year.
Formation et compétences requises :
You have a PhD, ideally in computer science or artificial intelligence.
You have competences in the foundations of machine learning methodologies (supervised, semi-supervised, unsupervised learning), knowledge and/or experience in modern deep learning paradigms (domain adaptation, transfer learning, weakly supervised learning and knowledge distillation), and
the development and exploitation of the latter in the context of signal processing and image analysis (computer vision).
Already proven experiences in the analysis of satellite and earth observation data with applications in the field of agriculture or environment would be appreciated.
Fluency in English is desirable, as well as long-term international experience: laureates who have not yet had such experience will be strongly encouraged to spend a period of time abroad, co-constructed with the host team, within 3 years after the internship year.
Adresse d’emploi :
500, rue Jean François Breton
34090 Montpellier, France