Post-doctoral offers 2023

Post-doctoral offers 2023

#DigitAg offers three 100% funded interdisciplinary post-doc grants

IMPORTANT: the applicant must not be:
- a former labeled, co-funded, or other PhD student from a #DigitAg member unit
- a former PhD student supervised/supervised by a #DigitAg associate researcher

The topics posted below have been selected and funded by #DigitAg
Want to know more? Get in touch with the contact person indicated for the topic you are interested in:

Engineering Sciences - Life and Environment Sciences – Social and Human Sciences

Development of a machine learning framework for the prediction of food security indicators at country scale from heterogeneous data

Keywords : Data Science, Machine Learning, Food Security, Heterogeneous Data

  • Contact: Roberto Interdonato - roberto.interdonato [AT] cirad.fr
  • Framing: Elodie Maitre D’Hotel, UMR Moisa, Cirad - Louis Reymondin, Alliance Bioversity International & CIAT, CGIAR
  • Units: Tetis – Moisa
  • #DigitAg: Axis 5: Data mining, data analysis, knowledge extraction, Axis 4: Information system, data storage and transfer, Challenge 0 : Cross-cutting topic, Challenge 6: Management of agricultural territories, Challenge 8: Agricultural development in the South

Food Security (FS) is a central problem in many areas of the world, as also testified by its presence as one of the 17 Sustainable Development Goals (SDG 2 - Zero Hunger). To monitor food insecurity situations, several early warning systems are active today, such as GIEWS (Global Information and Early Warning System, FAO),  and FEWSNET (Famine Early Warning Systems Network, USAID).
These systems use a limited set of data types, i.e., agroclimatic data from satellite images and indicators extracted from household surveys about nutritional, economical and production-related factors. Also, human intervention is often needed to combine and summarize all the sources of information.
Our hypothesis is that heterogeneous open data, related at different levels with food security, can be used to provide a machine learning based framework able to automatically produce FS indicators able to take into account the multiple and interrelated reasons behind this phenomenon. Some examples may be raster data containing spatial information, volunteered geographical information, meteorological data, quantitative economic indicators, and textual data from local news media.
The aim of this Post-Doc is to consolidate and extend the recent works on this topic resulting from a collaboration by UMR TETIS and UMR MOISA, mainly thanks to a PhD Thesis co-financed by #DigitAg.
More specifically, we want to develop a suite of user friendly data science methods (e.g., a Python library) able to retrieve and exploit heterogeneous data to effectively produce food security indicators at national scale.  Besides data processing in itself, the need to collect, integrate and assess the quality of such heterogeneous multi-source data introduces several additional challenges in this context.
The use of rich ground truth data provided by CGIAR (e.g., RHoMIS - Rural Household Multi-Indicator Survey) will allow to test the proposed methodologies on different countries in Africa and South-Eastern Asia.

 
Engineer Sciences - Life and Environment Sciences

High-throughput phenotyping of fruit tree genetic diversity for better adaptation to climate change
filled post-doc

Keywords: Digital phenotyping, genetic screening, fruit trees, transferability between species

  • Contact: Marie Weiss - marie.weiss [AT] inrae.fr
  • Framing: Marie Weiss, UMR Emmah, Inrae - Evelyne Costes, UMR Agap, Inrae
  • Units: Emmah – Agap
  • #DigitAg : Axis 2: Innovation in digital agriculture, Axis 3: Sensors and data acquisition and management, Challenge 2: Rapid phenotyping, Challenge 1: The agroecological challenge, Challenge 3: Crop protection

This project aims to characterize the resilience of fruit trees via digital phenotyping of a certain number of traits related to flowering and tree architecture based on the combination of the skills in image analysis (AGAP-PHENOMEN, EMMAH-CAPTE), genetics (GAFL) and architectural analysis (AFEF) of the four teams involved. In addition to methodological developments in stereovision and RGB data processing to assess these traits, this project is interested in the fusion of information generated by these data and the generalization and transferability of these traits and methods between species. Several approaches (machine and deep learning, mathematical morphology, statistics) will be combined. This project is based on the acquisition of datasets in three core collections (peach, apricot, apple) each of them comprising more than 150 different genotypes, thus ensuring a representativeness of the variability of flowering and tree structure by differences in age and contrasting environments. It also aims to establish temporal consistency between past measurements (visual notations) and digital phenotyping.
In terms of expected outputs, we aim to contribute to the phenotyping of complex and integrative traits (improved accuracy and throughput, access to new traits not accessible manually), and to the genetic screening of resilience by determining a typology of trees that can maintain production in the face of environmental perturbations.

 
Engineer Sciences - Life and Environment Sciences – Social and Human Sciences

Integration of heterogeneous data for simulation of livestock-wildlife contacts

Keywords : heterogeneous data, telemetry, remote sensing, spatial modelling, participatory modelling, extensive breeding, animal mobility

  • Contact: Tran Annelise - annelise.tran [AT] cirad.fr
  • Framing: Tran Annelise, UMR Tetis, Cirad – Le Page Christophe, UMR Sens, Cirad
  • Units: Tetis – Sens
  • #DigitAg: Axis 6: Modeling and simulation, Axis 3: Sensors and data acquisition and management, Challenge 6: Management of agricultural territories, Challenge 8: Agricultural development in the South

In many African regions, human population growth combined with the reduction of water resources due to climate change is leading to increased contacts, on the periphery of conservation areas, between rural communities living from agriculture and livestock and wildlife. In these interface systems, the mobilities of wild and domestic animals are source of events such as predation of livestock by carnivores, destruction of crops by wildlife, increased competition for shared natural resources, illegal hunting, and disease transmission. A better understanding of how climatic and environmental factors, as well as agricultural and livestock practices, determine these mobilities and potential contacts is a major challenge for identifying management methods to reconcile agricultural development and conservation issues.
The work of the proposed post-doctoral fellowship will be based on datasets from projects conducted in the periphery of Hwange National Park in Zimbabwe: i) telemetry data from GPS collars installed on wild and domestic ruminants to study their movements; ii) camera-traps images to measure the occurrence and frequency of inter-species contacts at watering holes; iii) time series of satellite images to monitor environmental changes at the interfaces of natural and agricultural areas; iv) herd management rules as described by the breeders. The objective will be to develop innovative methods for i) integrating these heterogeneous data to simulate the mobility and potential contacts of livestock and wildlife at the periphery of protected areas and ii) exploring scenarios defined in consultation with the various stakeholders in the territory.