A5088691547- Bayesian Methods and Mixture Models
- Statistical Methods and Inference
- Medical Image Segmentation Techniques
- Functional Brain Connectivity Studies
- Advanced MRI Techniques and Applications
- Gaussian Processes and Bayesian Inference
- Advanced Neuroimaging Techniques and Applications
- Brain Tumor Detection and Classification
- Gene expression and cancer classification
- Statistical Methods and Bayesian Inference
- Anomaly Detection Techniques and Applications
- Speech and Audio Processing
- Radiomics and Machine Learning in Medical Imaging
- Neural Networks and Applications
- Markov Chains and Monte Carlo Methods
- Medical Imaging Techniques and Applications
- Medical Imaging and Analysis
- Data-Driven Disease Surveillance
- Blind Source Separation Techniques
- Music and Audio Processing
- Financial Risk and Volatility Modeling
- Statistical Distribution Estimation and Applications
- AI in cancer detection
- Bayesian Modeling and Causal Inference
- Genetic and phenotypic traits in livestock
Laboratoire Jean Kuntzmann
2015-2025
Centre Inria de l'Université Grenoble Alpes
2016-2025
Institut national de recherche en informatique et en automatique
2012-2024
Institut polytechnique de Grenoble
2019-2024
Université Grenoble Alpes
2014-2024
Centre National de la Recherche Scientifique
2019-2024
Signify (Netherlands)
2023
ORCID
2020
Mathématiques, Informatique et Statistique pour l'Environnement et l'Agronomie
2016
Inserm
2014
The deviance information criterion (DIC) introduced by Spiegelhalter et al.(2002) for model assessment and comparison is directly inspired linear generalised models, but it open to different possible variations in the setting of missing data depending particular on whether or not variables are treated as parameters. In this paper, we reassess such models compare DIC constructions, testing behaviour these various extensions cases mixtures distributions random effect models.
Abstract On the basis of simulated data, this study compares relative performances Bayesian clustering computer programs structure , geneland geneclust and a new program named tess . While these four can detect population genetic from multilocus genotypes, only last three ones include simultaneous analysis geographical data. The are compared with respect to their abilities infer number populations, estimate membership probabilities, discontinuities clinal variation. results suggest that...
We present a study of multiple sclerosis segmentation algorithms conducted at the international MICCAI 2016 challenge. This challenge was operated using new open-science computing infrastructure. allowed for automatic and independent evaluation large range in fair completely manner. infrastructure used to evaluate thirteen methods MS lesions segmentation, exploring broad state-of-theart algorithms, against high-quality database 53 cases coming from four centers following common definition...
This paper addresses the issue of matching rigid and articulated shapes through probabilistic point registration. The problem is recast into a missing data framework where unknown correspondences are handled via mixture models. Adopting maximum likelihood principle, we introduce an innovative EM-like algorithm, namely, Expectation Conditional Maximization for Point Registration (ECMPR) algorithm. algorithm allows use general covariance matrices model components improves over isotropic case....
Data clustering has received a lot of attention and numerous methods, algorithms software packages are available. Among these techniques, parametric finite-mixture models play central role due to their interesting mathematical properties the existence maximum-likelihood estimators based on expectation-maximization (EM). In this paper we propose new mixture model that associates weight with each observed point. We introduce weighted-data Gaussian derive two EM algorithms. The first one...
Maximum likelihood estimation in finite mixture distributions is typically approached as an incomplete data problem to allow application of the expectation-maximization (EM) algorithm. In its general formulation, EM algorithm involves notion a complete space, which observed measurements and are embedded. An advantage that many difficult problems facilitated when viewed this way. One drawback simultaneous update used by standard requires overly informative spaces, leads slow convergence some...
In standard within-subject analyses of event-related functional magnetic resonance imaging (fMRI) data, two steps are usually performed separately: detection brain activity and estimation the hemodynamic response. Because these inherently linked, we adopt so-called region-based joint detection-estimation (JDE) framework that addresses this issue using a multivariate inference for estimation. JDE is built by making use regional bilinear generative model BOLD response constraining parameter...
Abstract Genomic offset statistics predict the maladaptation of populations to rapid habitat alteration based on association genotypes with environmental variation. Despite substantial evidence for empirical validity, genomic have well-identified limitations, and lack a theory that would facilitate interpretations predicted values. Here, we clarified theoretical relationships between unobserved fitness traits controlled by environmentally selected loci proposed geometric measure after change...
Hidden Markov random fields appear naturally in problems such as image segmentation, where an unknown class assignment has to be estimated from the observations at each pixel. Choosing probabilistic model that best accounts for is important first step quality of subsequent estimation and analysis. A commonly used selection criterion Bayesian Information Criterion (BIC) Schwarz (1978), but hidden fields, its exact computation not tractable due dependence structure induced by model. We propose...
The problem addressed in this paper is the automatic segmentation of stroke lesions on MR multi-sequences. Lesions enhance differently depending modality and there an obvious gain trying to account for various sources information a single procedure. To aim, we propose multimodal Markov random field model which includes all modalities simultaneously. results method proposed are compared with those obtained mono-dimensional applied each MRI sequence separately. We constructed Atlas blood...
Accurate tissue and structure segmentation of magnetic resonance (MR) brain scans is critical in several applications. In most approaches this task handled through two sequential steps. We propose to carry out cooperatively both subcortical by distributing a set local cooperative Markov random field (MRF) models. Tissue performed partitioning the volume into subvolumes where MRFs are estimated cooperation with their neighbors ensure consistency. Local estimation fits precisely intensity...
In this paper, we address the problems of modeling acoustic space generated by a full-spectrum sound source and using learned model for localization separation multiple sources that simultaneously emit sparse-spectrum sounds. We lay theoretical methodological grounds in order to introduce binaural manifold paradigm. perform an in-depth study latent low-dimensional structure high-dimensional interaural spectral data, based on corpus recorded with human-like audiomotor robot head. A nonlinear...
When analyzing brain tumors, two tasks are intrinsically linked, spatial localization, and physiological characterization of the lesioned tissues. Automated data-driven solutions exist, based on image segmentation techniques or parameters analysis, but for each task separately, other being performedmanually with user tuning operations. In this paper, availability quantitative magnetic resonance (MR) is combined advancedmultivariate statistical tools to design a fully automated method that...