PhD position OT-14635

DigitWelfare. A hybrid modelling approach to characterize dairy goat’s activity profiles associated with welfare

91120 Saclay

Back to jobs listing

INRAE presentation

The French National Research Institute for Agriculture, Food, and the Environment (INRAE) is a public research establishment. It is a community of 12,000 people with more than 200 research units and 42 experimental units located throughout France. The institute is among the world leaders in agricultural and food sciences, in plant and animal sciences, and is 11th in the world in ecology and environment. INRAE’s main goal is to be a key player in the transitions necessary to address major global challenges. In the face of the increase in population, climate change, scarcity of resources and decline in biodiversity, the institute develops solutions for multiperformance agriculture, high quality food and sustainable management of resources and ecosystems.

Work environment, missions and activities

Today, we all want animal production systems that are more respectful to animal welfare. Animal welfare is defined as a positive mental and physical state achieved by satisfaction of the animal’s physiological and behavioural needs according to its own perception of the situation (ANSES, 2018). Moreover, the agroecological transition of livestock farming integrates the concept of one welfare, where the interdependence between the welfare of animals, wellbeing of farmers and environmental protection is highlighted as part of this transition. To respond to these expectations, continuous objective measures of welfare state at the individual animal scale are needed. However, animal welfare is a multicriteria concept with different animal indicators combining behaviour, health, and aspects of performance and physiology. Such indicators are increasingly measurable with lower cost and less human effort thanks to PLF (Precision Livestock Farming) technology.

The objective of this thesis is to improve our ability to assess welfare by developing a pipeline for collection of heterogenous PLF (Precision Livestock Farming) data of performance and activity of dairy goats and develop a modelling approach (combination of supervised machine learning algorithm and dynamic model of perturbed production performance).  This will allow us to (1) determine the activity profiles associated with different husbandry practices considered in this project, and (2) to develop a multi-variable machine learning algorithm combined with dynamic model of goats’ performance considering behavioural and performance data, to detect deviations in the behavioural data in relation to perturbations of performance. The originality of our approach lies in the combining of the perturbed model of performance and a supervised machine learning algorithm. In particular, the following methods will be used: random forests (scikit-learn), gradient boosting with decision trees (xgboost, catboost, lightgbm), neural networks (tensorflow, keras). This project will break new ground in the characterization of activity profiles, and develop a novel modelling approach to detect deviations in these profiles. These deviations are potential signs of health and welfare problems.

This interdisciplinary thesis will benefit from the expertise of supervisors in data analytics, Artificial Intellegence, biology of robustness, and animal behaviour. This project will provide a rich research environment for the PhD candidate because it is associated with a number of projects and researcher networks in which the supervisorial team is implicated. The successful candidate will have the possibility to participate in the meetings and seminars of these networks. After the three years of the thesis, the PhD is expected to obtain a solid experience in interdisciplinary research.  The PhD student will acquire knowledge on what are the different types of modelling approaches and how the combination of such approaches enables valuable inferences to be extracted from existing data, in particular PLF data. She/he will be able to develop machine learning and concept driven models.  From an animal science point of view, the PhD candidate will become familiar with animal welfare concept and its influence on production sustainability. Most of all, she/he will gain the necessary tools and interdisciplinary experience to be able to handle such projects in the future as researcher or engineer.

The successful candidate will be integrated in the MoSAR team (systemic Modelling Applied to Ruminants) and the dynamic of doctoral school ABIES (Paris-Saclay university and AgroParisTech). She/he will have the possibility to follow graduate courses (English and French) of INRAE, university Paris-Saclay university and AgroParisTech.  

Training and skills

Master's degree/Engineering degree
  •         MSC (or diplôme d’ingenieur) in computer science, statistics, applied mathematics
  •         Knowledge on Python/R programming language
  •         Knowledge and experiences on the machine learning and modelling projects (Neural network, Random Forest, …)
  •         Experience in the domain of biology is advantageous
  •         High interest in animal welfare and behaviour
  •         Excellent communication skills to interact with a multi-disciplinary team.

INRAE's life quality

By joining our teams, you benefit from (depending on the type of contract and its duration):

- up to 30 days of annual leave + 15 days "Reduction of Working Time" (for a full time);
parenting support: CESU childcare, leisure services;
- skills development systems: trainingcareer advise;
social support: advice and listening, social assistance and loans;
holiday and leisure services: holiday vouchers, accommodation at preferential rates;
sports and cultural activities;
- collective catering.

How to apply

I send my CV and my motivation letter

Offer reference

  • Contract: PhD position
  • Duration: 3 years
  • Beginning: 01/07/2022
  • Reference: OT-14635
  • Dealine: 10/05/2022
Centre Ile-de-France - Versailles-Grignon

MOSAR 91120 Saclay


Living in France and working at INRAE Our guide for international scientists

Learn more