A5084617491- Statistical Methods and Inference
- Advanced Photocatalysis Techniques
- Bayesian Methods and Mixture Models
- Statistical Methods and Bayesian Inference
- TiO2 Photocatalysis and Solar Cells
- Graphene research and applications
- Stochastic processes and financial applications
- Fault Detection and Control Systems
- Transition Metal Oxide Nanomaterials
- Pigment Synthesis and Properties
- Markov Chains and Monte Carlo Methods
- Image and Signal Denoising Methods
- Statistical Methods in Clinical Trials
- Gaussian Processes and Bayesian Inference
- Water Quality Monitoring and Analysis
- Advanced Control Systems Optimization
- Graphene and Nanomaterials Applications
- Neural Networks and Applications
- Face and Expression Recognition
- Distributed Sensor Networks and Detection Algorithms
- Big Data and Business Intelligence
- Blind Source Separation Techniques
- Stochastic processes and statistical mechanics
- Advanced Statistical Methods and Models
- Mathematical Biology Tumor Growth
Vietnam National University Ho Chi Minh City
2020-2024
Ho Chi Minh City University of Science
2017-2024
Laboratoire de Mathématiques Raphaël Salem
2017-2021
Centre National de la Recherche Scientifique
2017-2020
Université de Rouen Normandie
2020
Université Paris-Sud
2017
Laboratoire de Mathématiques
2017
Université Paris Dauphine-PSL
2017
Université Paris Sciences et Lettres
2017
Laboratoire Paul Painlevé
2016-2017
Abstract We consider the statistical analysis of heterogeneous data for prediction, in situations where observations include functions, typically time series. extend modeling with mixtures-of-experts (ME), as a framework choice heterogeneity prediction vectorial observations, to this functional context. first present new family ME models, named (FME), which predictors are potentially noisy from entire functions. Furthermore, generating process predictor and real response, is governed by...
In the multidimensional setting, we consider errors-in- variables model. We aim at estimating unknown nonparametric multivariate regression function with errors in covariates. devise an adaptive estimators based on projection kernels wavelets and a deconvolution operator. propose automatic fully data driven procedure to select wavelet level resolution. obtain oracle inequality optimal rates of convergence over anisotropic Hölder classes. Our theoretical results are illustrated by some simulations.
We consider a stochastic individual-based model in continuous time to describe size-structured population for cell divisions. This is motivated by the detection of cellular aging biology. address here problem nonparametric estimation kernel ruling divisions based on eigenvalue related asymptotic behavior large population. inverse involves multiplicative deconvolution operator. Using Fourier technics we derive estimator whose consistency studied. The main difficulty comes from non-standard...
We consider a size-structured population describing the cell divisions. The is described by an empirical measure and we observe divisions in continuous time interval [0, T ]. address here problem of estimating division kernel h (or fragmentation kernel) case complete data. An adaptive estimator constructed based on function K with fully data-driven bandwidth selection method. obtain oracle inequality exponential convergence rate, for which optimality considered.
This paper focuses on estimating the invariant density function $f_X$ of strongly mixing stationary process $X_t$ in multiplicative measurement errors model $Y_t = X_t U_t$, where $U_t$ is also a process. We propose novel approach to handle non-independent data, typical real-world scenarios. For instance, data collected from various groups may exhibit interdependencies within each group, resembling generated $m$-dependent processes, subset processes. study extends applicability U_t$ diverse...
Abstract Parameters (mean and covariance matrix) estimation is often a problem of interest since it provides information about the location variation data correlation between features can be used for hypothesis testing, principle component analysis, etc. However, also common that values in some dataset are missing. A popular way to deal with this use an Expectation-Maximization algorithm or impute missing then estimate parameters based on imputed data. first approach local optimization may...
A two-class mixture model, where the density of one components is known, considered. We address issue nonparametric adaptive estimation unknown probability second component. propose a randomly weighted kernel estimator with fully data-driven bandwidth selection method, in spirit Goldenshluger and Lepski method. An oracle-type inequality for pointwise quadratic risk derived as well convergence rates over Holder smoothness classes. The theoretical results are illustrated by numerical simulations.
We consider the statistical analysis of heterogeneous data for prediction in situations where observations include functions, typically time series. extend modeling with Mixtures-of-Experts (ME), as a framework choice heterogeneity vectorial observations, to this functional context. first present new family ME models, named (FME) which predictors are potentially noisy from entire functions. Furthermore, generating process predictor and real response, is governed by hidden discrete variable...