This thesis addresses the problem of high dimensional inference.We propose different methods for estimating normalizing constants and sampling complex distributions.In a first part, we develop several Markov chain Monte Carlo methods.On the one hand, we develop a new approach for non-reversible kernels. On the other hand, we propose two massively parallelizable methods combining local and global properties of Markov chain Monte Carlo methods, in particular based on a new normalization constant estimator.We apply these methods to the approximate inference of the posterior distribution of deep Bayesian neural networks, in a case where the state space is very high dimensional.In a second part, we propose two generative models, based on a new f...
Automatic decision making and pattern recognition under uncertainty are difficult tasks that are ubi...
Bayesian statistics has emerged as a leading paradigm for the analysis of complicated datasets and f...
Computational Bayesian statistics builds approximations to the posterior distribution either bysampl...
This thesis addresses the problem of high dimensional inference.We propose different methods for est...
This thesis addresses the problem of high dimensional inference.We propose different methods for est...
Cette thèse porte sur le problème de l'inférence en grande dimension.Nous proposons différentes mét...
Monte Carlo methods are are an ubiquitous tool in modern statistics. Under the Bayesian paradigm, th...
We propose a new computationally efficient sampling scheme for Bayesian inference involving high dim...
We propose a new computationally efficient sampling scheme for Bayesian inference involving high dim...
The availability of massive computational resources has led to a wide-spread application and develop...
The availability of massive computational resources has led to a wide-spread application and develop...
National audienceBayesian posterior distributions can be numerically intractable, even by the means ...
We present a simple new Monte Carlo algorithm for evaluating probabilities of observations in comple...
Automatic decision making and pattern recognition under uncertainty are difficult tasks that are ubi...
Computational Bayesian statistics builds approximations to the posterior distribution either bysampl...
Automatic decision making and pattern recognition under uncertainty are difficult tasks that are ubi...
Bayesian statistics has emerged as a leading paradigm for the analysis of complicated datasets and f...
Computational Bayesian statistics builds approximations to the posterior distribution either bysampl...
This thesis addresses the problem of high dimensional inference.We propose different methods for est...
This thesis addresses the problem of high dimensional inference.We propose different methods for est...
Cette thèse porte sur le problème de l'inférence en grande dimension.Nous proposons différentes mét...
Monte Carlo methods are are an ubiquitous tool in modern statistics. Under the Bayesian paradigm, th...
We propose a new computationally efficient sampling scheme for Bayesian inference involving high dim...
We propose a new computationally efficient sampling scheme for Bayesian inference involving high dim...
The availability of massive computational resources has led to a wide-spread application and develop...
The availability of massive computational resources has led to a wide-spread application and develop...
National audienceBayesian posterior distributions can be numerically intractable, even by the means ...
We present a simple new Monte Carlo algorithm for evaluating probabilities of observations in comple...
Automatic decision making and pattern recognition under uncertainty are difficult tasks that are ubi...
Computational Bayesian statistics builds approximations to the posterior distribution either bysampl...
Automatic decision making and pattern recognition under uncertainty are difficult tasks that are ubi...
Bayesian statistics has emerged as a leading paradigm for the analysis of complicated datasets and f...
Computational Bayesian statistics builds approximations to the posterior distribution either bysampl...