Post-doctoral project (2009-2010) in the Mét@risk Research Unit, Modeling and Statistics Team, Charanpal Dhanjal

This project aims at developing tools for the automatic regognition of a phenotype based on metabolomic data, the methodology is founded on recent concepts borrowed from the machine learning field (e.g. optimization of the ROC curve, «multi-class» ranking, functional data analysis).

Post-doctoral project (2006-2008) in the Mét@risk Research Unit, Modeling and Statistics Team, Anne-Laure Afchain

A new approach is proposed and new tools are being developed to study the emergence of a pathogenic bacterium in the food chain – the case of B. cereus in non-sterile products undergoing thermal treatment, in the context of a National Research Program on Food and Human Nutrition (Programme National de Recherches en Alimentation et Nutrition Humaine, PNRA) and National Research Agency (Agence Nationale de la Recherche, ANR) project.
Manager of the Work package on "Modeling of the risk of emergence of B. cereus in the food chain".

Post-doctoral project (2007-2008) in the Mét@risk Research Unit, Modeling and Statistics Team, Mélanie Zetlaoui

The overall problem consists of assessing exposure to food-borne risks in individuals. In this context, this post-doctoral project focuses on breaking down the long-term weekly household purchasing data (SECODIP) available in order to predict individual levels of consumption. Statistical learning techniques (Lasso, Ridge, CART, SVM...) are applied to tackle this problem in a supervised approach: the decomposition phenomenon is learnt from another source of data providing individual levels of consumption for all family members, but only over a single week (INCA).

This post-doctoral project forms part of the ANR White Project TAMIS.