Suchergebnisse

National audience ; La question de l'échantillonnage optimal d'information spatiale a été largement étudiée dans le cas d'observations à valeurs réelles. A l'inverse, peu de travaux traitent du cas d'observations discrètes, pourtant pertinant dans de nombreux problèmes de cartographie de systèmes biologiques. Nous proposons d'exploiter le cadre des champs de Markov, classiquement utilisés en analyse d'image, à la fois pour représenter la distribution spatiale du phéomène étudié, et également pour définir une notion d'utilisé d'un échantillon. Nous considérons le problème de la conception d'une stratégie adaptative d'éhantillonnage et l'échantillon optimal est défini comme celui maximisant un critère d'utilité en espérance quantifiant la qualité de la carte reconstruire et les coûts d'échantillonnage. Nous présentons ensuite un algorithme de résolution de type apprentissage par renforcement, reposant sur la modélisation du problème d'origine dans le cadre des processus décisionnels de Markov.En nous appuyant sur une validation empirique, nous illustrons les avantages de cette approche qui apporte un gain significatif en temps de calcul et fournit des politiques approchées satisfaisantes.

International audience ; AbstractThere is a growing demand for moving towards sustainable agri-food systems which per nature covers a complex network of activities and domains; such systems will benefit from multi-criteria decision analysis (MCDA) methods. Although some reviews on MCDA in agri-food research have been published, none of them covered the whole value chain. In this article, a corpus of 954 articles published by INRA scientists from 2007 to 2017 was used to study the diversity and potentiality of MCDA techniques. For the first time, experts from more than 10 agri-food domains worked altogether to annotate the articles, carry out a multivariate analysis, and finally interpret the statistical results to identify the specificities of certain domains and the complementarities between domains and to suggest avenues for future agri-food research. One-third of the studies were based only on a list of indicators, even when their purpose was to choose, sort, or rank options. Regardless of the scientific discipline in the agri-food sector, MCDA studies rarely considered temporal dynamics, spatial scale changes, or stakeholder contributions. As the agri-food system becomes increasingly sustainable in the near future, the use of MCDA methods will accelerate. To become more effective, they will have to include ecosystem services, even outside the scope of ecological studies. Similarly, MCDA studies will need to include participatory science to involve stakeholders (i.e., public authorities, governmental agencies) and end-users (i.e., farmers, producers, industrials, consumers) in the construction of the multi-criteria evaluation but also in the resulting decisions.

International audience ; AbstractThere is a growing demand for moving towards sustainable agri-food systems which per nature covers a complex network of activities and domains; such systems will benefit from multi-criteria decision analysis (MCDA) methods. Although some reviews on MCDA in agri-food research have been published, none of them covered the whole value chain. In this article, a corpus of 954 articles published by INRA scientists from 2007 to 2017 was used to study the diversity and potentiality of MCDA techniques. For the first time, experts from more than 10 agri-food domains worked altogether to annotate the articles, carry out a multivariate analysis, and finally interpret the statistical results to identify the specificities of certain domains and the complementarities between domains and to suggest avenues for future agri-food research. One-third of the studies were based only on a list of indicators, even when their purpose was to choose, sort, or rank options. Regardless of the scientific discipline in the agri-food sector, MCDA studies rarely considered temporal dynamics, spatial scale changes, or stakeholder contributions. As the agri-food system becomes increasingly sustainable in the near future, the use of MCDA methods will accelerate. To become more effective, they will have to include ecosystem services, even outside the scope of ecological studies. Similarly, MCDA studies will need to include participatory science to involve stakeholders (i.e., public authorities, governmental agencies) and end-users (i.e., farmers, producers, industrials, consumers) in the construction of the multi-criteria evaluation but also in the resulting decisions.

International audience ; AbstractThere is a growing demand for moving towards sustainable agri-food systems which per nature covers a complex network of activities and domains; such systems will benefit from multi-criteria decision analysis (MCDA) methods. Although some reviews on MCDA in agri-food research have been published, none of them covered the whole value chain. In this article, a corpus of 954 articles published by INRA scientists from 2007 to 2017 was used to study the diversity and potentiality of MCDA techniques. For the first time, experts from more than 10 agri-food domains worked altogether to annotate the articles, carry out a multivariate analysis, and finally interpret the statistical results to identify the specificities of certain domains and the complementarities between domains and to suggest avenues for future agri-food research. One-third of the studies were based only on a list of indicators, even when their purpose was to choose, sort, or rank options. Regardless of the scientific discipline in the agri-food sector, MCDA studies rarely considered temporal dynamics, spatial scale changes, or stakeholder contributions. As the agri-food system becomes increasingly sustainable in the near future, the use of MCDA methods will accelerate. To become more effective, they will have to include ecosystem services, even outside the scope of ecological studies. Similarly, MCDA studies will need to include participatory science to involve stakeholders (i.e., public authorities, governmental agencies) and end-users (i.e., farmers, producers, industrials, consumers) in the construction of the multi-criteria evaluation but also in the resulting decisions.

International audience ; AbstractThere is a growing demand for moving towards sustainable agri-food systems which per nature covers a complex network of activities and domains; such systems will benefit from multi-criteria decision analysis (MCDA) methods. Although some reviews on MCDA in agri-food research have been published, none of them covered the whole value chain. In this article, a corpus of 954 articles published by INRA scientists from 2007 to 2017 was used to study the diversity and potentiality of MCDA techniques. For the first time, experts from more than 10 agri-food domains worked altogether to annotate the articles, carry out a multivariate analysis, and finally interpret the statistical results to identify the specificities of certain domains and the complementarities between domains and to suggest avenues for future agri-food research. One-third of the studies were based only on a list of indicators, even when their purpose was to choose, sort, or rank options. Regardless of the scientific discipline in the agri-food sector, MCDA studies rarely considered temporal dynamics, spatial scale changes, or stakeholder contributions. As the agri-food system becomes increasingly sustainable in the near future, the use of MCDA methods will accelerate. To become more effective, they will have to include ecosystem services, even outside the scope of ecological studies. Similarly, MCDA studies will need to include participatory science to involve stakeholders (i.e., public authorities, governmental agencies) and end-users (i.e., farmers, producers, industrials, consumers) in the construction of the multi-criteria evaluation but also in the resulting decisions.

International audience ; AbstractThere is a growing demand for moving towards sustainable agri-food systems which per nature covers a complex network of activities and domains; such systems will benefit from multi-criteria decision analysis (MCDA) methods. Although some reviews on MCDA in agri-food research have been published, none of them covered the whole value chain. In this article, a corpus of 954 articles published by INRA scientists from 2007 to 2017 was used to study the diversity and potentiality of MCDA techniques. For the first time, experts from more than 10 agri-food domains worked altogether to annotate the articles, carry out a multivariate analysis, and finally interpret the statistical results to identify the specificities of certain domains and the complementarities between domains and to suggest avenues for future agri-food research. One-third of the studies were based only on a list of indicators, even when their purpose was to choose, sort, or rank options. Regardless of the scientific discipline in the agri-food sector, MCDA studies rarely considered temporal dynamics, spatial scale changes, or stakeholder contributions. As the agri-food system becomes increasingly sustainable in the near future, the use of MCDA methods will accelerate. To become more effective, they will have to include ecosystem services, even outside the scope of ecological studies. Similarly, MCDA studies will need to include participatory science to involve stakeholders (i.e., public authorities, governmental agencies) and end-users (i.e., farmers, producers, industrials, consumers) in the construction of the multi-criteria evaluation but also in the resulting decisions.

International audience ; AbstractThere is a growing demand for moving towards sustainable agri-food systems which per nature covers a complex network of activities and domains; such systems will benefit from multi-criteria decision analysis (MCDA) methods. Although some reviews on MCDA in agri-food research have been published, none of them covered the whole value chain. In this article, a corpus of 954 articles published by INRA scientists from 2007 to 2017 was used to study the diversity and potentiality of MCDA techniques. For the first time, experts from more than 10 agri-food domains worked altogether to annotate the articles, carry out a multivariate analysis, and finally interpret the statistical results to identify the specificities of certain domains and the complementarities between domains and to suggest avenues for future agri-food research. One-third of the studies were based only on a list of indicators, even when their purpose was to choose, sort, or rank options. Regardless of the scientific discipline in the agri-food sector, MCDA studies rarely considered temporal dynamics, spatial scale changes, or stakeholder contributions. As the agri-food system becomes increasingly sustainable in the near future, the use of MCDA methods will accelerate. To become more effective, they will have to include ecosystem services, even outside the scope of ecological studies. Similarly, MCDA studies will need to include participatory science to involve stakeholders (i.e., public authorities, governmental agencies) and end-users (i.e., farmers, producers, industrials, consumers) in the construction of the multi-criteria evaluation but also in the resulting decisions.