4LMBA - Laboratoire de Mathématiques de Bretagne Atlantique (Université de Brest, 6 avenue Le Gorgeu, CS 93837, 29238 Brest cedex 3 / Université de Bretagne-Sud, Centre Yves Coppens, Bât. B, Campus de Tohannic, BP 573, 56017 Vannes - France)
Abstract : This article proposes a method to model signals of oysters' openings over time using a four-state renewal process. Two of them are of particular interest and correspond to instants when the animals are open or closed. An estimator of the cumulative jump rate of the renewal process is provided. It relies on observations of the jumps between the four states. Here these measures are not available but the observed signal is assumed to take ranges of real values according to this underlying process. A procedure to estimate a probability density function that summarizes the information of the signal is explained. This leads to estimation of the hidden renewal process and of its cumulative jump rate for each oyster. We propose to classify these estimated functions for a group of oysters in order to discriminate these animals according to their health status. Such a diagnosis is essential when using these animals as biosensors for water quality assessment.
https://hal.archives-ouvertes.fr/hal-00958616
Contributeur : Gilles Durrieu
<>
Soumis le : jeudi 13 mars 2014 - 04:03:59
Dernière modification le : jeudi 29 novembre 2018 - 16:53:05