Browsing
     by title


0-9 A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

or enter first few letters:   
OK
Full Text
See detailBaulicher Brandschutz
Schneider, Ulrich; Franssen, Jean-Marc ULg; Lebeda, Christian

Book published by Bauwerk Verlag, GmbH - 2nd edition (2008)

This book contains: fire safety concepts as a base for fire safety of building constructions; fire safety requirements ; etc

Detailed reference viewed: 102 (10 ULg)
Full Text
Peer Reviewed
See detailA Bayesian Approach for Modeling Origin-Destination Matrices
Perrakis, Konstantinos; Karlis, Dimitris; Cools, Mario ULg et al

in Proceedings of the 90th Annual Meeting of the Transportation Research Board (DVD-ROM) (2011)

The majority of Origin Destination (OD) matrix estimation methods focus on situations where weak or partial information, derived from sample travel surveys, is available. Information derived from travel ... [more ▼]

The majority of Origin Destination (OD) matrix estimation methods focus on situations where weak or partial information, derived from sample travel surveys, is available. Information derived from travel census studies, in contrast, covers the entire population of a specific study area of interest. In such cases where reliable historical data exist, statistical methodology may serve as a flexible alternative to traditional travel demand models by incorporating estimation of trip-generation, trip-attraction and trip-distribution in one model. In this research, a statistical Bayesian approach on OD matrix estimation is presented, where modeling of OD flows, derived from census data, is related only to a set of general explanatory variables. The assumptions of a Poisson model and of a Negative-Binomial model are investigated on a realistic application area concerning the region of Flanders on the level of municipalities. Problems related to the absence of closed-form expressions are bypassed with the use of a Markov Chain Monte Carlo algorithm, known as the Metropolis-Hastings algorithm. Additionally, a strategy is proposed in order to obtain predictions from the hierarchical, Poisson-Gamma structure of the Negative-Binomial model conditional on the posterior expectations of the mixing parameters. In general, Bayesian methodology reduces the overall uncertainty of the estimates by delivering posterior distributions for the parameters of scientific interest as well as predictive distributions for future OD flows. Predictive goodness-of-fit tests suggest a good fit to the data and overall results indicate that the approach is applicable on large networks, with relatively low computational and explanatory data-gathering costs. [less ▲]

Detailed reference viewed: 16 (0 ULg)
Full Text
Peer Reviewed
See detailA Bayesian approach for modeling origin–destination matrices
Perrakis, Konstantinos; Karlis, Dimitris; Cools, Mario ULg et al

in Transportation Research. Part A : Policy & Practice (2012), 46(1), 200212

The majority of origin destination (OD) matrix estimation methods focus on situations where weak or partial information, derived from sample travel surveys, is available. Information derived from travel ... [more ▼]

The majority of origin destination (OD) matrix estimation methods focus on situations where weak or partial information, derived from sample travel surveys, is available. Information derived from travel census studies, in contrast, covers the entire population of a specific study area of interest. In such cases where reliable historical data exist, statistical methodology may serve as a flexible alternative to traditional travel demand models by incorporating estimation of trip-generation, trip-attraction and trip-distribution in one model. In this research, a statistical Bayesian approach on OD matrix estimation is presented, where modeling of OD flows derived from census data, is related only to a set of general explanatory variables. A Poisson and a negative binomial model are formulated in detail, while emphasis is placed on the hierarchical Poisson-gamma structure of the latter. Problems related to the absence of closed-form expressions are bypassed with the use of a Markov Chain Monte Carlo method known as the Metropolis–Hastings algorithm. The methodology is tested on a realistic application area concerning the Belgian region of Flanders on the level of municipalities. Model comparison indicates that negative binomial likelihood is a more suitable distributional assumption than Poisson likelihood, due to the great degree of overdispersion present in OD flows. Finally, several predictive goodness-of-fit tests on the negative binomial model suggest a good overall fit to the data. In general, Bayesian methodology reduces the overall uncertainty of the estimates by delivering posterior distributions for the parameters of scientific interest as well as predictive distributions for future OD flows. [less ▲]

Detailed reference viewed: 18 (5 ULg)
Full Text
See detailBayesian approach to integrate molecular data into genetic evaluations
Gengler, Nicolas ULg; Verkenne, Catherine

in Interbull Bulletin (2007), 37

Detailed reference viewed: 8 (1 ULg)
Full Text
Peer Reviewed
See detailBayesian data fusion applied to water table spatial mapping
Fasbender, D.; Peeters, Luk; Bogaert, P. et al

in Water Resources Research (2008), 44

Water table elevations are usually sampled in space using piezometric measurements that are unfortunately expensive to obtain and are thus scarce over space. Most of the time, piezometric data are ... [more ▼]

Water table elevations are usually sampled in space using piezometric measurements that are unfortunately expensive to obtain and are thus scarce over space. Most of the time, piezometric data are sparsely distributed over large areas, thus providing limited direct information about the level of the corresponding water table. As a consequence, there is a real need for approaches that are able at the same time to (1) provide spatial predictions at unsampled locations and (2) enable the user to account for all potentially available secondary information sources that are in some way related to water table elevations. In this paper, a recently developed Bayesian data fusion (BDF) framework is applied to the problem of water table spatial mapping. After a brief presentation of the underlying theory, specific assumptions are made and discussed to account for a digital elevation model and for the geometry of a corresponding river network. On the basis of a data set for the Dijle basin in the north part of Belgium, the suggested model is then implemented and results are compared to those of standard techniques such as ordinary kriging and cokriging. Respective accuracies and precisions of these estimators are finally evaluated using a ‘‘leave-one-out’’ cross-validation procedure. Although the BDF methodology was illustrated here for the integration of only two secondary information sources (namely, a digital elevation model and the geometry of a river network), the method can be applied for incorporating an arbitrary number of secondary information sources, thus opening new avenues for the important topic of data integration in a spatial mapping context. [less ▲]

Detailed reference viewed: 146 (7 ULg)
Full Text
Peer Reviewed
See detailBayesian Data Fusion for water table interpolation: incorporating a hydrogeological conceptual model in kriging
Peeters, Luk; Fasbender, Dominique; Batelaan, Okke et al

in Water Resources Research (2010), 46(8), 08532

The creation of a contour map of the water table in an unconfined aquifer based on head measurements is often the first step in any hydrogeological study. Geostatistical interpolation methods (e.g ... [more ▼]

The creation of a contour map of the water table in an unconfined aquifer based on head measurements is often the first step in any hydrogeological study. Geostatistical interpolation methods (e.g. kriging) may provide exact interpolated groundwater levels at the measurement locations, but often fail to represent the hydrogeological flow system. A physically based, numerical groundwater model with spatially variable parameters and inputs is more adequate in representing a flow system. Due to the difficulty in parameterization and solving the inverse problem however, an often considerable difference between calculated and observed heads will remain. In this study the water table interpolation methodology presented by Fasbender et al. (2008), in which the results of a kriging interpolation are combined with information from a drainage network and a Digital Elevation Model (DEM), using the Bayesian Data Fusion framework (Bogaert and Fasbender, 2007), is extended to incorporate information from a tuned analytic element groundwater model. The resulting interpolation is exact at the measurement locations while the shape of the head contours is in accordance with the conceptual information incorporated in the groundwater flow model. The Bayesian Data Fusion methodology is applied to a regional, unconfined aquifer in Central Belgium. A cross-validation procedure shows that the predictive capability of the interpolation at unmeasured locations benefits from the Bayesian Data Fusion of the three data sources (kriging, DEM and groundwater model), compared to the individual data sources or any combination of two data sources. [less ▲]

Detailed reference viewed: 107 (13 ULg)
Full Text
Peer Reviewed
See detailBayesian density estimation from grouped continuous data
Lambert, Philippe ULg; Eilers, Paul H.C.

in Computational Statistics & Data Analysis (2009), 53

Grouped data occur frequently in practice, either because of limited resolution of instruments, or because data have been summarized in relatively wide bins. A combination of the composite link model with ... [more ▼]

Grouped data occur frequently in practice, either because of limited resolution of instruments, or because data have been summarized in relatively wide bins. A combination of the composite link model with roughness penalties is proposed to estimate smooth densities from such data in a Bayesian framework. A simulation study is used to evaluate the performances of the strategy in the estimation of a density, of its quantiles and rst moments. Two illustrations are presented: the rst one involves grouped data of lead concentrations in the blood and the second one the number of deaths due to tuberculosis in The Netherlands in wide age classes. [less ▲]

Detailed reference viewed: 29 (10 ULg)
Full Text
See detailBayesian Design Space applied to Pharmaceutical Development
Lebrun, Pierre ULg

Doctoral thesis (2012)

Given the guidelines such as the Q8 document published by the International Conference on Harmonization (ICH), that describe the “Quality by Design” paradigm for the Pharmaceutical Development, the aim of ... [more ▼]

Given the guidelines such as the Q8 document published by the International Conference on Harmonization (ICH), that describe the “Quality by Design” paradigm for the Pharmaceutical Development, the aim of this work is to provide a complete methodology addressing this problematic. As a result, various Design Spaces were obtained for different analytical methods and a manufacturing process. In Q8, Design Space has been defined as the “the multidimensional combination and interaction of input variables (e.g., material attributes) and process parameters that have been demonstrated to provide assurance of quality” for the analytical outputs or processes involved in Pharmaceutical Development. Q8 is thus clearly devoted to optimization strategies and robustness studies. In the beginning of this work, it was noted that existing statistical methodolo- gies in optimization context were limited as the predictive framework is based on mean response predictions. In such situations, the data and model uncertainties are generally completely ignored. This often leads to increase the risks of taking wrong decision or obtaining unreliable manufactured product. The reasons why it happens are also unidentified. The “assurance of quality” is clearly not addressed in this case. To improve the predictive nature of statistical models, the Bayesian statistical framework was used to facilitate the identification of the predictive distribution of new outputs, using numerical simulations or mathematical derivations when possi- ble. By use of the improved models in a risk-based environment, separation analytical methods such as the high performance liquid chromatography were studied. First, optimal solutions of separation of several compounds in mixtures were identified. Second, the robustness of the methods was simultaneously assessed thanks to the risk-based Design Space identification. The usefulness of the methodology was also demonstrated in the optimization of the separation of subsets of relevant compounds, without additional experiments. The high guarantee of quality of the optimized methods allowed easing their use for their very purpose, i.e., the tracing of compounds and their quantification. Transfer of robust methods to high-end equipments was also simplified. In parallel, one sub-objective was the total automation of analytical method de- velopment and validation. Some data treatments including the Independent Com- ponent Analysis and clustering methodologies were found more than promising to provide accurate automated results. Next, the Design Space methodology was applied to a small-scale spray-dryer manufacturing process. It also allowed the expression of guarantees about the quality of the obtained powder. Finally, other predictive models including mixed-effects models were used for the validation of analytical and bio-analytical quantitative methods. [less ▲]

Detailed reference viewed: 61 (12 ULg)
Full Text
Peer Reviewed
See detailA Bayesian Design Space for analytical methods based on multivariate models and predictions
Lebrun, Pierre ULg; Boulanger, Bruno ULg; Debrus, Benjamin ULg et al

in Journal of Biopharmaceutical Statistics (2013), 23

The International Conference for Harmonization (ICH) has released regulatory guidelines for Pharmaceutical Development. In the document ICH Q8, The Design Space of a process is presented as the set of ... [more ▼]

The International Conference for Harmonization (ICH) has released regulatory guidelines for Pharmaceutical Development. In the document ICH Q8, The Design Space of a process is presented as the set of factor settings providing satisfactory results. However, ICH Q8 does not propose any practical methodology to define, derive and compute Design Space. In parallel, in the last decades, it has been observed that the diversity and the quality of analytical methods have evolved exponentially allowing substantial gains in selectivity and sensitivity. However, there is still a lack for a rationale towards the development of robust separation methods in a systematic way. Applying ICH Q8 to analytical methods provides a methodology for predicting a region of the space of factors in which results will be reliable. Combining design of experiments and Bayesian standard multivariate regression, an identified form of the predictive distribution of a new response vector has been identified and used, under non-informative as well as informative prior distributions of the parameters. From the responses and their predictive distribution, various critical quality attributes can be easily derived. This Bayesian framework was then extended to the multi-criteria setting to estimate the predictive probability that several critical quality attributes will be jointly achieved in the future use of an analytical method. An example based on a high-performance liquid chromatography (HPLC) method is given. For this example, a constrained sampling scheme was applied to ensure the modeled responses have desirable properties. [less ▲]

Detailed reference viewed: 177 (58 ULg)
Full Text
Peer Reviewed
See detailBayesian estimation of evoked and induced responses
Friston, Kark J.; Henson, Richard; Phillips, Christophe ULg et al

in Human Brain Mapping (2006), 27(9), 722-735

We describe an extension of our empirical Bayes approach to magnetoencephalography/electroencephalography (MEG/EEG) source reconstruction that covers both evoked and induced responses. The estimation ... [more ▼]

We describe an extension of our empirical Bayes approach to magnetoencephalography/electroencephalography (MEG/EEG) source reconstruction that covers both evoked and induced responses. The estimation scheme is based on classical covariance component estimation using restricted maximum likelihood (ReML). We have focused previously on the estimation of spatial covariance components under simple assumptions about the temporal correlations. Here we extend the scheme using temporal basis functions to place constraints on the temporal form of the responses. We show how the same scheme can estimate evoked responses that are phase-locked to the stimulus and induced responses that are not. For a single trial the model is exactly the same, In the context of multiple trials, however, the inherent distinction between evoked and induced responses calls for different treatments of the underlying hierarchical multitrial model. We derive the respective models and show how they can be estimated efficiently using ReML. This enables the Bayesian estimation of evoked and induced changes in power or, more generally, the energy of wavelet coefficients. [less ▲]

Detailed reference viewed: 7 (0 ULg)
Full Text
Peer Reviewed
See detailBayesian estimation of the true prevalence, sensitivity and specificity of the Rose Bengal and indirect ELISA tests in the diagnosis of bovine brucellosis.
Sanogo, Moussa; Thys, Eric; Achi, Yaba L. et al

in Veterinary Journal (2013), 195

Serology is the most convenient method for detecting brucellosis but the efficient use of such tests in disease control requires evaluation of diagnostic performance and discriminative ability. The ... [more ▼]

Serology is the most convenient method for detecting brucellosis but the efficient use of such tests in disease control requires evaluation of diagnostic performance and discriminative ability. The objective of this study was to assess the performance of the Rose Bengal test (RBT) and an indirect ELISA (iELISA) in diagnosing brucellosis in 995 serum samples collected from cattle in the Ivory Coast between 2005 and 2009. A Bayesian approach was used to evaluate the two tests by estimating their sensitivities and specificities. The correlation-adjusted sensitivity of the iELISA was estimated to be 96.1% (credibility interval [CrI], 92.7-99.8), whereas that of the RBT was 54.9% (CrI, 23.5-95.1). High correlation-adjusted specificities were found for both tests (95.0%; [CrI, 91.1-99.6] for the iELISA and 97.7%; [CrI, 95.3-99.4] for the RBT, respectively). The true prevalence of brucellosis was estimated from the serum samples to be 4.6% (95%; [CrI, 0.6-9.5]). The level of agreement between the two tests was evaluated using indices of agreement (n=995). Good agreement was found for negative results (96.6%; confidence interval [CI], 95.7-97.4), a finding supported by an estimated significant correlation of 0.37 (95%; CI, 0.01-0.73) within the sera testing negative. Agreement was lower for sera testing positive (52.2% CI: 41.9-62.5). The findings highlight the importance of using these two tests in combination as part of any brucellosis control programme. [less ▲]

Detailed reference viewed: 12 (2 ULg)
Full Text
Peer Reviewed
See detailA Bayesian framework for the ratio of two Poisson rates in the context of vaccine efficacy trials
Laurent, Stéphane ULg; Legrand, Catherine

in ESAIM: Probability and Statistics = Probabilité et statistique : P & S (2011)

In many applications, we assume that two random observations x and y are generated according to independent Poisson distributions and we are interested in performing statistical inference on the ratio of ... [more ▼]

In many applications, we assume that two random observations x and y are generated according to independent Poisson distributions and we are interested in performing statistical inference on the ratio of the two incidence rates, called the relative risk in vaccine efficacy trials, in which context x and y are the numbers of cases in the vaccine and the control groups respectively. In this paper we start by defining a natural semi-conjugate family of prior distributions for this model, allowing straightforward computation of the posterior inference. Following theory on reference priors, we define the reference prior for the partial immunity model when the relative risk is the parameter of interest. We also define a family of reference priors with partial information on the incidence rate of the unvaccinated population while remaining uninformative about the relative risk . We notice that these priors belong to the semi-conjugate family. We then demonstrate using numerical examples that Bayesian credible intervals enjoy attractive frequentist properties when using reference priors, a typical property of reference priors. [less ▲]

Detailed reference viewed: 82 (26 ULg)
Peer Reviewed
See detailBayesian hierarchical linear regression for the validation of analytical methods
Lebrun, Pierre ULg; Boulanger, Bruno ULg; Rozet, Eric ULg et al

Conference (2010, May)

Analytical quantitative methods are widely used to quantify analytes of interest, for instance in pharmaceutical formulations, linking an observed response to the concentration of one compound of the ... [more ▼]

Analytical quantitative methods are widely used to quantify analytes of interest, for instance in pharmaceutical formulations, linking an observed response to the concentration of one compound of the formulation. Current methodologies to validate these analytical methods are based on one-way ANOVA random effect model in order to estimate repeatability and intermediate precision variances. This model is then applied several times at different concentration levels over a range of concentrations where the method is intended to be used, assuming independency between the levels. In this way, the capacity of the method to be able to quantify accurately is assessed at various concentration levels, and the method is said to be fitted for purpose (or valid) at the concentration level(s) where it shows trueness and precision that are fully acceptable, i.e. within predefined acceptance limits. Problem of such approach is the amount of data required and the time needed to collect them, while small sample sizes (small number of series and of replicates per series) are often preferred and practiced by laboratories. A better use of the data could then be envisaged. In this presentation, we take into account the response-concentration relationship that exists by the use of a hierarchical linear regression model. Instead of fitting a model at each concentration level that is assessed, only one model is studied. We show how the Bayesian framework is well adapted to this task. Also, as a predictive tool, we naturally derive beta-expectation and beta-gamma content tolerance intervals by means of MCMC simulations. The Bayesian modeling can also include informative prior information whenever justified, leading to reliable decisions given the domain knowledge. [less ▲]

Detailed reference viewed: 88 (9 ULg)
Full Text
See detailBayesian integration of external information into the single step approach for genomically enhanced prediction of breeding values
Vandenplas, Jérémie ULg; Misztal, Ignacy; Faux, Pierre ULg et al

Conference (2012, July 17)

An assumption to compute unbiased estimated breeding values (EBV) is that all information, i.e. genomic, pedigree and phenotypic information, has to be considered simultaneously. However, current ... [more ▼]

An assumption to compute unbiased estimated breeding values (EBV) is that all information, i.e. genomic, pedigree and phenotypic information, has to be considered simultaneously. However, current developments of genomic selection will bias evaluations because only records related to selected animals will be available. The single step genomic evaluation (ssGBLUP) could reduce pre-selection bias by the combination of genomic, pedigree and phenotypic information which are internal for the ssGBLUP. But, in opposition to multi-step methods, external information, i.e. information from outside ssGBLUP, like EBV and associated reliabilities from Multiple Across Country Evaluation which represent a priori known phenotypic information, are not yet integrated into the ssGBLUP. To avoid multi-step methods, the aim of the study was to assess the potential of a Bayesian procedure to integrate a priori known external information into a ssGBLUP by considering simplifications of computational burden, a correct propagation of external information and no multiple considerations of contributions due to relationships. To test the procedure, 2 dairy cattle populations (referenced by “internal” and “external”) were simulated as well as milk production for the first lactation of each female in both populations. Internal females were randomly mated with internal and 50 external males. Genotypes of 3000 single-nucleotide polymorphisms for the 50 males were simulated. A ssGBLUP was applied as the internal evaluation. The external evaluation was based on phenotypic and pedigree external information. External information integrated into the ssGBLUP consisted to external EBV and associated reliabilities of the 50 males. Results showed that rank correlations among Bayesian EBV and EBV based on the joint use of external and internal data and genomic information were higher than 0.99 for the 50 males and internal animals. The respective correlations for the internal evaluation were equal to 0.50 and 0.90. Thereby, the Bayesian procedure can integrate external information into ssGBLUP. [less ▲]

Detailed reference viewed: 61 (9 ULg)
Full Text
Peer Reviewed
See detailBayesian integration of external information into the single step approach for genomically enhanced prediction of breeding values
Vandenplas, Jérémie ULg; Misztal, Ignacy; Faux, Pierre ULg et al

in Journal of Dairy Science (2012), 95(Supplement 2),

Detailed reference viewed: 31 (4 ULg)
See detailBayesian inversion for induced responses
Mattout, Jérémie; Phillips, Christophe ULg; Daunizeau, Jean et al

in Friston, Karl; Ashburner, John; Kiebel, Stefan (Eds.) et al Statistical Parametric Mapping: the analysis of functional brain images (2007)

The book Statistical Parametric Mapping: The Analysis of Functional Brain Images (2007) provides the background and methodology for the analysis of all types of brain imaging data, from functional ... [more ▼]

The book Statistical Parametric Mapping: The Analysis of Functional Brain Images (2007) provides the background and methodology for the analysis of all types of brain imaging data, from functional magnetic resonance imaging to magnetoencephalography. Critically, "Statistical Parametric Mapping" provides a widely accepted conceptual framework which allows treatment of all these different modalities. The book takes the reader from the basic concepts underlying the analysis of neuroimaging data to cutting edge approaches that would be difficult to find in any other source. The material is presented in an incremental way so that the reader can understand the precedents for each new development. This book will be particularly useful to neuroscientists engaged in any form of brain mapping; who have to contend with the real-world problems of data analysis and understanding the techniques they are using. It is primarily a scientific treatment and a didactic introduction to the analysis of brain imaging data. It can be used as both a textbook for students and scientists starting to use the techniques, as well as a reference for practicing neuroscientists. [less ▲]

Detailed reference viewed: 10 (0 ULg)
See detailBayesian inversion of EEG models
Mattout, Jérémie; Phillips, Christophe ULg; Daunizeau, Jean et al

in Friston, Karl; Ashburner, John; Kiebel, Stefan (Eds.) et al Statistical Parametric Mapping: the analysis of functional brain images (2007)

The book Statistical Parametric Mapping: The Analysis of Functional Brain Images (2007) provides the background and methodology for the analysis of all types of brain imaging data, from functional ... [more ▼]

The book Statistical Parametric Mapping: The Analysis of Functional Brain Images (2007) provides the background and methodology for the analysis of all types of brain imaging data, from functional magnetic resonance imaging to magnetoencephalography. Critically, "Statistical Parametric Mapping" provides a widely accepted conceptual framework which allows treatment of all these different modalities. The book takes the reader from the basic concepts underlying the analysis of neuroimaging data to cutting edge approaches that would be difficult to find in any other source. The material is presented in an incremental way so that the reader can understand the precedents for each new development. This book will be particularly useful to neuroscientists engaged in any form of brain mapping; who have to contend with the real-world problems of data analysis and understanding the techniques they are using. It is primarily a scientific treatment and a didactic introduction to the analysis of brain imaging data. It can be used as both a textbook for students and scientists starting to use the techniques, as well as a reference for practicing neuroscientists. [less ▲]

Detailed reference viewed: 11 (0 ULg)
Full Text
Peer Reviewed
See detailBayesian methods for predicting and modelling winter wheat biomass
Mansouri, Majdi ULg; Dumont, Benjamin ULg; Destain, Marie-France ULg

Poster (2014, February)

The objectives of this paper are threefold. The first objective is to propose to use an Improved Particle Filtering (IPF) based on minimizing Kullback-Leibler divergence for crop models' predictions. The ... [more ▼]

The objectives of this paper are threefold. The first objective is to propose to use an Improved Particle Filtering (IPF) based on minimizing Kullback-Leibler divergence for crop models' predictions. The performances of the proposed technique are compared with those of the conventional Particle Filtering (PF) for improving nonlinear crop model predictions. The main novelty of this task is to develop a Bayesian algorithm for nonlinear and non-Gaussian state and parameter estimation with better proposal distribution. The second objective is to investigate the effects of practical challenges on the performances of state estimation algorithms PF and IPF. Such practical challenges include (i) the effect of measurement noise on the estimation performances and (ii) the number of states and parameters to be estimated. The third objective is to use the state estimation techniques PF and IPF for updating prediction of nonlinear crop model in order to predict winter wheat biomass. PF and IPF are applied at a dynamic crop model with the aim to predict a state variable, namely the winter wheat biomass, and to estimate several model parameters. Furthermore, the effect of measurement noise (e.g., different signal-to-noise ratios) on the performances of PF and IPF is investigated. The results of the comparative studies show that the IPF provides a significant improvement over the PF because, unlike the PF which depends on the choice of sampling distribution used to estimate the posterior distribution, the IPF yields an optimum choice of the sampling distribution, which also accounts for the observed data. [less ▲]

Detailed reference viewed: 11 (2 ULg)
Full Text
Peer Reviewed
See detailBayesian methods for predicting LAI and soil moisture
Mansouri, Majdi ULg; Dumont, Benjamin ULg; Destain, Marie-France ULg

in Proceedings of the 11th International Conference on Precision Agriculture (2012)

The prediction errors of crop models are often important due to uncertainties in the estimates of initial values of the states, in parameters, and in equations. The measurements needed to run the model ... [more ▼]

The prediction errors of crop models are often important due to uncertainties in the estimates of initial values of the states, in parameters, and in equations. The measurements needed to run the model are sometimes not numerous or known with some uncertainty. In this paper, two Bayesian filtering methods were used to update the state variable values predicted by MiniSTICS model. The chosen state variates were the LAI (Leaf Area Index) of a wheat crop (Triticum aestivum L.) and the corresponding moisture content of two soil layers (0-20 cm and 30-50 cm). These state variates were estimated simultaneously with several parameters. The assessed filtering methods were the centralized Particle Filtering (PF) and the Variational Bayesian Filtering (VF). The former is known to be sensitive to the number of particles while the latter yields an optimal choice of the sampling distribution over the state variable by minimizing the Kullback-Leibler divergence. In fact, variational calculus leads to a simple Gaussian sampling distribution whose parameters (estimated iteratively) depends on the observed data. On basis of a case study, the VF method was found more efficient than the PF method. Indeed, with the VF, the Root Mean Square Error (RMSE) of the three estimated states was smaller and the convergence of the all parameters was ensured. [less ▲]

Detailed reference viewed: 116 (18 ULg)