Éditeur(s) : HAL CCSD Ecological Society of America
Résumé : Although niche-based and stochastic processes, including dispersal limitation and demographic stochasticity, can each contribute to community assembly, it is difficult to quantify the relative importance of each process in natural vegetation. Here, we extend Shipley's maxent model (Community Assembly by Trait Selection, CATS) for the prediction of relative abundances to incorporate both trait-based filtering and dispersal limitation from the larger landscape and develop a statistical decomposition of the proportions of the total information content of relative abundances in local communities that are attributable to trait-based filtering, dispersal limitation, and demographic stochasticity. We apply the method to tree communities in a mature, species-rich, tropical forest in French Guiana at 1-, 0.25- and 0.04-ha scales. Trait data consisted of species' means of 17 functional traits measured over both the entire meta-community and separately in each of nine 1-ha plots. Trait means calculated separately for each site always gave better predictions. There was clear evidence of trait-based filtering at all spatial scales. Trait-based filtering was the most important process at the 1-ha scale (34%), whereas demographic stochasticity was the most important at smaller scales (37-53%). Dispersal limitation from the meta-community was less important and approximately constant across scales (similar to 9%), and there was also an unresolved association between site-specific traits and meta-community relative abundances. Our method allows one to quantify the relative importance of local niche-based and meta-community processes and demographic stochasticity during community assembly across spatial and temporal scales.
ISSN: 0012-9658

hal-01032424
https://hal.archives-ouvertes.fr/hal-01032424

Éditeur(s) : HAL CCSD
Résumé : 27 pages
Cascade Stochastic Differential Equation (SDE), a continuous time model for energy dissipation in turbulence, is a generalization of the Yaglom discrete cascade model. We extend this SDE to a model in random environment by assuming that its two parameters are switched by a continuous time Markov chain whose states represent the states of the environment. Moreover, a Dirichlet process is placed as a prior on the space of sample paths of this chain. We propose a Bayesian estimation method of this model which is tested both on simulated data and on real data of wind speed measured at the entrance of the mangrove ecosystem in Guadeloupe.
https://hal.archives-ouvertes.fr/hal-00286131

hal-00286131
https://hal.archives-ouvertes.fr/hal-00286131
https://hal.archives-ouvertes.fr/hal-00286131/document
https://hal.archives-ouvertes.fr/hal-00286131/file/Srik11_PhysicaDDspace.pdf

Éditeur(s) : HAL CCSD
Résumé : A standard approach for large scale image classification involves high dimensional features and Stochastic Gradient Descent algorithm (SGD) for the minimization of classical Hinge Loss in the primal space. Although complexity of Stochastic Gradient Descent is linear with the number of samples these method suffers from slow convergence. In order to cope with this issue, we propose here a Stochastic Low-Rank Newton Descent SLND for minimization of any calibrated loss in the primal space. SLND approximates the inverse Hessian by the best low-rank approximation according to squared Frobenius norm. We provide core optimization for fast convergence. Theoretically speaking, we show explicit convergence rates of the algorithm using these calibrated losses, which in addition provide working sets of parameters for experiments. Experiments are provided on the SUN, Caltech256 and ImageNet databases, with simple, uniform and efficient ways to tune remaining SLND parameters. On each of these databases, SLND challenges the accuracy of SGD with a speed of convergence faster by order of magnitude.
https://hal.archives-ouvertes.fr/hal-00825414

hal-00825414
https://hal.archives-ouvertes.fr/hal-00825414
https://hal.archives-ouvertes.fr/hal-00825414/document
https://hal.archives-ouvertes.fr/hal-00825414/file/TechnicalReport.pdf

Éditeur(s) : Université des Antilles et de la Guyane
Résumé : The need for an integrated approach to fisheries is now widely affirmed, including the FAO, in particular in the context of tropical artisanal fisheries. These small-scale fisheries, often multispecies are very important in terms of production, employment and food security. However these fisheries and the exploited resources are often characterized by a lack of data making difficult their management within an ecosystem approach, and by the ecological and economic system complexities including trophic and technological interactions. This thesis aims to contribute to the development of bioeconomic tools for small tropical fisheries in a sustainable management perspective based on the ecosystem approach. In this perspective, the French Guiana coastal fishery constitutes an interesting case study. The thesis proposes multi-criteria evaluations, complex bioeconomic models and viable management scenarios for this fishery. At first, multivariate statistical analysis suggests a satisfactory overall status of the fishery in terms of sustainability. However, some performance differences are noted within the fishery, particularly at border areas. Also, management improvements are proposed. Then, through bioeconomic modeling, projections of different fishing scenarios show that, in the long-term, the current exploitation level may not be consistent with the future increase of local demand and a loss of biodiversity may occur. A scenario called "co-viability" reconciling ecological, economic and social objectives, with a high probability of achievement is exhibited. Finally, the comparison of optimal behavior in cooperative and non-cooperative conditions, confirms that harvest levels are greater when actors do not cooperate. Furthermore, it is shown that the state of the ecosystem depends on the fishing strategies and the type of interaction between species. Beyond the diagnosis made for the case study, the method is promising in the context of small tropical fisheries, while the co-viability approach allows finding the exploitation conditions under which ecological and socio-economic sustainability are meet, what the conventional fisheries management generally do not allow.
La nécessité d’une approche intégrée des pêches est actuellement largement affirmée, notamment par la FAO, en particulier dans le contexte des pêcheries artisanales tropicales. Ces pêcheries à petites échelles, souvent multi-spécifiques sont très importantes en termes d’emploi et de production, y compris pour la sécurité alimentaire. Néanmoins ces pêcheries et la biodiversité exploitées sont souvent marquées, d’une part, par le manque de données rendant difficile leur gestion dans le cadre d’une approche écosystémique, d’autre part, par la complexité des systèmes écologiques et économiques sous-jacents incluant interactions trophiques et techniques. Cette thèse contribue à la mise au point d’outils bioéconomiques adaptés aux petites pêcheries tropicales dans la perspective d’une gestion durable des pêches fondée sur l’approche écosystémique. Dans cette perspective, la pêcherie côtière en Guyane Française constitue un cas d’étude particulièrement fécond. La thèse propose ainsi des évaluations multi-critères, des modèles bioéconomiques complexes et des scénarios de gestion viable pour cette pêcherie. Dans un premier temps l’analyse statistique multivariée suggère un statut global satisfaisant de la pêcherie en termes de durabilité. Cependant des différences de performance sont notées au sein de la pêcherie, notamment au niveau des zones frontalières. Aussi des améliorations dans le mode de gestion sont proposées. Ensuite, à travers la modélisation bioéconomique, les projections des différents scénarios de pêche montrent qu’à long terme le niveau d’exploitation actuel peut ne pas être en adéquation avec la future augmentation de la demande locale et qu’une perte de biodiversité peut avoir lieu. Un scenario dit de "co-viabilité" conciliant des objectifs à la fois écologique, économique et social, avec une probabilité de réalisation satisfaisante, est mis en exergue. Enfin, la comparaison des comportements optimaux en situation coopératif et non coopératif, montre dans quelle mesure la viabilité est favorisée quand les acteurs coopèrent. Au-delà du diagnostic apporté pour le cas d’étude, la méthode utilisée s’avère prometteuse dans le contexte des petites pêcheries tropicales, tandis que l’approche de co-viabilité permet de trouver les modalités d’exploitation dans le cadre de compromis entre durabilité écologique et socio-économique, ce que les principes de gestion halieutiques traditionnels ne permettent généralement pas.

Droits : 2013 Université des Antilles et de la Guyane
http://archimer.ifremer.fr/doc/00144/25492/23646.pdf
http://archimer.ifremer.fr/doc/00144/25492/

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
The objectives are to study and model the aggregate wind power fluctuations dynamics in the multifractal framework. We present here the analysis of aggregate power output sampled at 1 Hz during three years. We decompose the data into several Intrinsic Mode Functions (IMF) using Empirical Mode Decomposition (EMD). We use a new approach, arbitrary order Hilbert spectral analysis, a combination of EMD approach with Hilbert spectral analysis (or Hilbert Huang Transform) and the classical structure function analysis to extract the scaling exponents or multifractal spectrum $zeta(q)$: this function provides a full characterization of a process at all intensities and all scales. The application of both methods, i.e. structure function and arbitrary-order Hilbert spectral analyses, give similar results indicating that the aggregate power output from a wind farm, possesses intermittent and multifractal properties. In order to check this result, we generate stochastic simulations of a Multifractal Random Walk (MRW) using a log-normal stochastic equation. We show that the simulation results are fully compatible with the experimental results.
Physica A

hal-01527887
https://hal.archives-ouvertes.fr/hal-01527887
DOI : 10.1016/j.physa.2013.04.038

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
Joint inversions are now commonly used in the earth sciences. They have been developed to better understand the structure of the earth, since they provide more constraints on the inverted parameters. We propose a new process to simultaneously invert several data sets in order to better image 3D crustal and upper mantle structures. Our inversion uses three kinds of data that present good complementarity: (1) P-wave receiver functions to provide Moho depth variations, (2) teleseismic delay times of P-waves to retrieve velocity anomalies in the crust and the upper mantle, and (3) gravity anomalies to image density variations at the lithospheric scale. We use a stochastic scheme, where receiver functions are first inverted. The interpolated resulting Moho depths are incorporated as a priori information into the joint inversion of teleseismic delay times and gravity anomalies process. Moreover, velocity and density can be linked by empirical relationships, which justifies the joint inversion of those parameters. In our stochastic approach, we perform a model space search for Moho variations, P-velocity, and density structure to find an acceptable fit to the three data sets. In order to preferentially sample the good data fit regions, we chose the neighborhood algorithm of Sambridge to optimistically survey the model space. We model the delay times with 3D raytracing using evenly spaced velocity-density nodes. We present here the first results given by this method on synthetic tests. (C) 2011 Elsevier Ltd. All rights reserved.
ISSN: 0098-3004

hal-00643158
https://hal.archives-ouvertes.fr/hal-00643158
DOI : 10.1016/j.cageo.2010.11.017

Éditeur(s) : HAL CCSD
Résumé : Point processes are often used as spatial or spatio-temporal distribution models of occurrences. In this PhD dissertation, we focus first on Cox processes driven by a hidden process associated with a Dirichlet process. This model corresponds to hidden occurrences influencing the stochastic intensity of observed occurrences. We generalize the notion of "Shot noise Cox process" introduced by M{o}ller and develop its bayesian analysis by a Gibbs sampler combined with a Metropolis-Hastings algorithm. We show that our MCMC method is a reversible jump one. The model takes into account a random number of hidden contributions producing effects on the observed point process intensity. Therefore the parameter space has a variable dimension. We focus the statistical inference on the estimation of the hidden contribution expected value, the hidden contribution expected number, the spatial influence and correlation parameters. The contribution equality test and contribution independence test are proposed. Applications in epidemiology and ecology are shown from extit{Rubus fruticosa} data, extit{Ibicella lutea} data and death number data in counties of Georgia, USA. Two situations are considered with respect to available data~: firstly, the spatial positions of occurrences are observed between several pairs of consecutive dates; secondly, counts are carried out over a fixed time interval in several spatial sampling units. Secondly, we focus on point processes with memory introduced by Kagan, Ogata and Vere-Jones, pioneers in statistical seismology. In fact, spatio-temporal point processes play an important role in the studies of earthquake catalogs since they consist of seismic events with their dates and spatial locations. We studied an ETAS (Epidemic Type Aftershock Sequence) model with time independent background intensity and several triggering functions taking into account previous events. We illustrate our approach with a seismicity study of the Lesser Antilles arc. A comparaison study of Gamma, Weibull, Log-Normal and modified Omori law triggering function models is also carried out. We show that the modified Omori law does not fit the Lesser Antilles seismic data and the best adjusted triggering function is the Weibull model. Consequently, the waiting time between aftershocks is weaker in the Lesser Antilles zone compared to the one in regions with seismicity described by the modified Omori law. In other words, aftershock aggregativity is higher in the Lesser Antilles region. The stochastic background intensity following a Dirichlet process centered on a spatial log-normal process is discussed.
Les processus ponctuels sont souvent utilisés comme modèles de répartitions spatiales ou spatio-temporelles d'occurrences. Dans cette thèse, nous nous intéressons tout d'abord à des processus de Cox dirigés par un processus caché associé à un processus de Dirichlet. Ce modèle correspond à des occurrences cachées influençant l'intensité stochastique des occurrences observées. Nous généralisons la notion de " Shot noise Cox process " introduite par Moller et développons le traitement bayésien par un échantillonneur de Gibbs combiné à un algorithme de Metropolis-Hastings. Nous montrons que cette méthode MCMC est à sauts réversibles. Le modèle prend en compte, en effet, un nombre aléatoire de contributions cachées influençant l'intensité du processus ponctuel observé donc a un espace paramétrique de dimension variable. Nous focalisons l'inférence statistique sur l'estimation de la valeur espérée de chaque contribution cachée, le nombre espéré de contributions cachées, le degré d'influence spatiale de ces contributions et leur degré de corrélation. Le test d'égalité des contributions et celui de leur indépendance sont ainsi développés. L'utilité en épidémiologie et en écologie est alors démontrée à partir de données de Rubus fruticosa, Ibicella lutea et de mortalité dans les cantons de Georgia, USA. En termes de données observées, deux situations sont considérées: premièrement, les positions spatiales des occurrences sont observées entre plusieurs paires de dates consécutives; deuxièmement, des comptages sont effectués, au cours d'une période fixée, dans des unités d'échantillonnage spatiales. D'autre part, nous nous intéressons aux processus ponctuels à mémoire introduits par Kagan, Ogata et Vere-Jones, précurseurs de la statistique sismologique. En effet, les processus ponctuels spatio-temporels ont une place importante dans l'étude des catalogues sismiques puisque ces derniers sont généralement constitués d'événements sismiques datés et géo-référencés. Nous avons étudié un modèle ETAS (Epidemic Type Aftershock Sequence) avec une intensité d'arrière-plan indépendante du temps et plusieurs fonctions déclenchantes permettant d'intégrer les événements antérieurs récents. Cette approche est utilisée pour étudier la sismicité de l'arc des Petites Antilles. Une étude comparative des modèles Gamma, Weibull, Log-Normal et loi d'Omori modifiée pour les fonctions déclenchantes est menée. Nous montrons que la loi d'Omori modifiée ne s'ajuste pas aux données sismiques des Petites Antilles et la fonction déclenchante la plus adaptée est le modèle de Weibull. Cela implique que le temps d'attente entre répliques dans la zone des Petites Antilles est plus faible que celui des régions à sismicité décrite par la loi d'Omori modifiée. Autrement dit, l'agrégation des répliques après un événement majeur est plus prononcée dans la zone des Petites Antilles. La possibilité d'inclure une intensité d'arrière-plan suivant un processus de Dirichlet centré sur un processus spatial log-gaussien est discutée.
https://tel.archives-ouvertes.fr/tel-00841146

Droits : info:eu-repo/semantics/OpenAccess
NNT : 2012AGUY0555
tel-00841146
https://tel.archives-ouvertes.fr/tel-00841146
https://tel.archives-ouvertes.fr/tel-00841146/document
https://tel.archives-ouvertes.fr/tel-00841146/file/these_VALMY.pdf

Éditeur(s) : HAL CCSD Springer Verlag/EDP Sciences
Résumé : No efficient method is available to compare multi-locus estimates of diversity while taking into account inter-locus and inter-population stochastic variance. The advent of genome scan approaches makes the development of such tests absolutely necessary. We developed a method to compare genome-wide diversity estimates while taking into account-and factoring out-variation in census size and making use of inter-locus variance to assess significance of differences in diversity levels. An approach based on rarefaction with bootstrap re-sampling (RaBoT) was implemented into a test of multi-locus comparison of diversity coded in R. The properties of the test were studied by applying it to simulated populations with varying diversity levels and varying differences in diversity levels. The test was then applied to empirical data from disturbed and undisturbed populations of Virola michelii (Myristicaceae) genotyped at 693 amplified fragment length polymorphism (AFLP) markers. RaBoT was found to be rather conservative, with large numbers of false negatives when the diversity in the compared populations was similar, and false positives mostly associated to comparisons of populations with extremely high levels of diversity. When applied to empirical data, RaBoT detected higher genetic diversity in a post-disturbance than in an undisturbed population and lower genetic diversity in a seedling than in the corresponding adult population, but it also revealed differences in diversity between subgroups within the disturbed and undisturbed plots. RaBoT is a sensitive method to compare multi-locus levels of diversity that can be applied both at the genotype level for dominant markers (e.g. AFLP) and at the allele level for biallelic codominant markers (e.g. single-nucleotide polymorphisms).
ISSN: 1286-4560

hal-01032413
https://hal.archives-ouvertes.fr/hal-01032413
DOI : 10.1007/s13595-013-0302-z

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
The objectives are to study and model the aggregate wind power fluctuations dynamics in the multifractal framework. We present here the analysis of aggregate power output sampled at 1 Hz during three years. We decompose the data into several Intrinsic Mode Functions (IMFs) using Empirical Mode Decomposition (EMD). We use a new approach, arbitrary order Hilbert spectral analysis, a combination of the EMD approach with Hilbert spectral analysis (or Hilbert-Huang Transform) and the classical structure-function analysis to extract the scaling exponents or multifractal spectrum ζ(q)ζ(q): this function provides a full characterization of a process at all intensities and all scales. The application of both methods, i.e. structure-function and arbitrary-order Hilbert spectral analyses, gives similar results indicating that the aggregate power output from a wind farm, possesses intermittent and multifractal properties. In order to check this result, we generate stochastic simulations of a Multifractal Random Walk (MRW) using a log-normal stochastic equation. We show that the simulation results are fully compatible with the experimental results
Physica A

hal-00846643
https://hal.archives-ouvertes.fr/hal-00846643
DOI : 10.1016/j.physa.2013.04.038

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
The quasi-equilibrium approximation is acceptable when molecular interactions are fast enough compared to circuit dynamics, but is no longer allowed when cellular activities are governed by rare events. A typical example is the lactose operon (lac), one of the most famous paradigms of transcription regulation, for which several theories still coexist to describe its behaviors. The lac system is generally analyzed by using equilibrium constants, contradicting single-event hypotheses long suggested by Novick and Weiner (1957). Enzyme induction as an all-or-none phenomenon. Proc. Natl. Acad. Sci. USA 43, 553-566) and recently refined in the study of (Choi et al., 2008. A stochastic single-molecule event triggers phenotype switching of a bacterial cell. Science 322, 442-446). In the present report, a lac repressor (LacI)-mediated DNA immunoprecipitation experiment reveals that the natural LacI-lac DNA complex built in vivo is extremely tight and long-lived compared to the time scale of lac expression dynamics, which could functionally disconnect the abortive expression bursts and forbid using the standard modes of lac bistability. As alternatives, purely kinetic mechanisms are examined for their capacity to restrict induction through: (i) widely scattered derepression related to the arrival time variance of a predominantly backward asymmetric random walk and (ii) an induction threshold arising in a single window of derepression without recourse to nonlinear multimeric binding and Hill functions. Considering the complete disengagement of the lac repressor from the lac promoter as the probabilistic consequence of a transient stepwise mechanism, is sufficient to explain the sigmoidal lac responses as functions of time and of inducer concentration. This sigmoidal shape can be misleadingly interpreted as a phenomenon of equilibrium cooperativity classically used to explain bistability, but which has been reported to be weak in this system.
ISSN: 0022-5193

inserm-00818506
http://www.hal.inserm.fr/inserm-00818506
http://www.hal.inserm.fr/inserm-00818506/document
http://www.hal.inserm.fr/inserm-00818506/file/LacI-TeX.pdf
ARXIV : 1312.5566

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
Recently, a large amount of work has been devoted to the study of spectral clustering — a powerful unsupervised classification method. This paper brings contributions to both its foundations, and its applications to text classification. Departing from the mainstream, concerned with hard membership, we study the extension of spectral clustering to soft membership (probabilistic, EM style) assignments. One of its key features is to avoid the complexity gap of hard membership. We apply this theory to a challenging problem, text clustering for languages having permeable borders, via a novel construction of Markov chains from corpora. Experiments with a readily available code clearly display the potential of the method, which brings a visually appealing soft distinction of languages that may define altogether a whole corpus.
ISSN: 0031-3203

hal-00329991
https://hal.archives-ouvertes.fr/hal-00329991
DOI : 10.1016/j.patcog.2008.06.024

Éditeur(s) : HAL CCSD Springer Verlag (Germany)
Résumé : International audience
The Michaelis-Menten enzymatic reaction is sufficient to perceive many subtleties of network modeling, including the concentration and time scales separations, the formal equivalence between bulk phase and single-molecule approaches, or the relationships between single-cycle transient probabilities and steady state rates. Twelve methods proposed by different authors and yielding the same famous Michaelis-Menten equation, are selected here to illustrate the kinetic and probabilistic use of rate constants and to review basic techniques for handling them. Finally, the general rate of an ordered multistep reaction, of which the Michaelis-Menten reaction is a particular case, is deduced from a Markovian approach.
ISSN: 0259-9791

hal-00920806
https://hal-univ-rennes1.archives-ouvertes.fr/hal-00920806
https://hal-univ-rennes1.archives-ouvertes.fr/hal-00920806v2/document
https://hal-univ-rennes1.archives-ouvertes.fr/hal-00920806/file/12MM.pdf
ARXIV : 1312.5468

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
We study diffusion in a heterogeneous medium that is characterized by spatially varying diffusion properties from a random walk point of view. We show that an inhomogeneous continuous time random walk (CTRW) with a spatially variable exponential transition time distribution solves the spatially discretized heterogeneous diffusion equation. This demonstrates the equivalence of the widely used time-domain random walk (TDRW) scheme and spatially inhomogeneous CTRW and at the same time provides a demonstration of the formal equivalence of the TDRW particle formulation and the heterogeneous diffusion equation. Based on this equivalence, we develop a TDRW method for heterogeneous diffusion under spatially variable multirate mass transfer properties. We discuss the implementation of these schemes and study the diffusion behavior in the presence of traps that are characterized by a truncated power-law trapping time distribution.
ISSN: 0309-1708

hal-00767566
https://hal.archives-ouvertes.fr/hal-00767566
DOI : 10.1016/j.advwatres.2012.07.015

Éditeur(s) : HAL CCSD Royal Society, The
Résumé : International audience
A 30-year study of temporal changes in gastropod community structure on the reefs of a Pacific Ocean atoll (Fangataufa, Tuamotu Archipelago) subjected to atmospheric nuclear tests during the 1960s offered the opportunity for an otherwise impossible field experiment that could help ecologists understand mollusc primary succession. Reef molluscs were partly or entirely wiped out by the heat of the nuclear tests and the reefs were recolonized by ocean larvae. On all reefs, community composition before the tests was very different from what it evolved to afterwards. A new method of analysis was developed to study the temporal variation in community composition before versus after the tests (temporal beta diversity). Analyses showed that community compositions diverged through time among the reefs. Only some species can survive the harsh conditions of supralittoral zones, so the same species recolonized them; environmental filtering controlled the development of the new communities. In the reef flat and edge zones, differences in community composition seem to be the result of neutral stochastic colonization by larvae coming from the open ocean. All reefs developed a community composition quite different from that before the nuclear tests.
ISSN: 0962-8452

hal-01174175
https://hal-univ-perp.archives-ouvertes.fr/hal-01174175
DOI : 10.1098/rspb.2015.0750

Éditeur(s) : HAL CCSD Springer Verlag (Germany)
Résumé : International audience
The Michaelis-Menten enzymatic reaction is sufficient to perceive many subtleties of network modeling, including the concentration and time scales separations, the formal equivalence between bulk phase and single-molecule approaches, or the relationships between single-cycle transient probabilities and steady state rates. Seven methods proposed by different authors and yielding the same famous Michaelis-Menten equation, are selected here to illustrate the kinetic and probabilistic use of rate constants and to review basic techniques for handling them. Finally, the general rate of an ordered multistep reaction, of which the Michaelis-Menten reaction is a particular case, is deduced from a Markovian approach.
Journal of Mathematical Chemistry

hal-00920806
https://hal-univ-rennes1.archives-ouvertes.fr/hal-00920806
https://hal-univ-rennes1.archives-ouvertes.fr/hal-00920806/document
ARXIV : 1312.5468
DOI : 10.1007/s10910-013-0237-5