24 pages, including 2 pages of references and 10 pages of appendixIn machine learning, it is common to optimize the parameters of a probabilistic model, modulated by a somewhat ad hoc regularization term that penalizes some values of the parameters. Regularization terms appear naturally in Variational Inference (VI), a tractable way to approximate Bayesian posteriors: the loss to optimize contains a Kullback--Leibler divergence term between the approximate posterior and a Bayesian prior. We fully characterize which regularizers can arise this way, and provide a systematic way to compute the corresponding prior. This viewpoint also provides a prediction for useful values of the regularization factor in neural networks. We apply this framewor...
This paper is concerned with the construction of prior probability measures for parametric families ...
Recent works have investigated deep learning models trained by optimising PAC-Bayes bounds, with pri...
There are two major routes to address linear inverse problems. Whereas regularization-based approach...
24 pages, including 2 pages of references and 10 pages of appendixIn machine learning, it is common ...
International audienceWe discuss a long-lasting {\em qui pro quo} between regularization-based and B...
The companion technical report hal-00862925 contains additional worked examplesInternational audienc...
International audienceWe investigate deep Bayesian neural networks with Gaussian priors on the weigh...
Deep neural networks have bested notable benchmarks across computer vision, reinforcement learning, ...
We propose a Bayesian framework for regression problems, which covers areas which are usually dealt ...
Throughout the last decade, deep learning has reached a sufficient level of maturity to become the p...
AbstractIn this paper we show how a user can influence recovery of Bayesian networks from a database...
We develop some new error bounds for learning algorithms induced by regularization methods in the re...
In linear regression problems with many predictors, penalized regression techniques are often used t...
Bayesian predictive inference provides a coherent description of entire predictive uncertainty throu...
Variational inference with a factorized Gaussian posterior estimate is a widely-used approach for le...
This paper is concerned with the construction of prior probability measures for parametric families ...
Recent works have investigated deep learning models trained by optimising PAC-Bayes bounds, with pri...
There are two major routes to address linear inverse problems. Whereas regularization-based approach...
24 pages, including 2 pages of references and 10 pages of appendixIn machine learning, it is common ...
International audienceWe discuss a long-lasting {\em qui pro quo} between regularization-based and B...
The companion technical report hal-00862925 contains additional worked examplesInternational audienc...
International audienceWe investigate deep Bayesian neural networks with Gaussian priors on the weigh...
Deep neural networks have bested notable benchmarks across computer vision, reinforcement learning, ...
We propose a Bayesian framework for regression problems, which covers areas which are usually dealt ...
Throughout the last decade, deep learning has reached a sufficient level of maturity to become the p...
AbstractIn this paper we show how a user can influence recovery of Bayesian networks from a database...
We develop some new error bounds for learning algorithms induced by regularization methods in the re...
In linear regression problems with many predictors, penalized regression techniques are often used t...
Bayesian predictive inference provides a coherent description of entire predictive uncertainty throu...
Variational inference with a factorized Gaussian posterior estimate is a widely-used approach for le...
This paper is concerned with the construction of prior probability measures for parametric families ...
Recent works have investigated deep learning models trained by optimising PAC-Bayes bounds, with pri...
There are two major routes to address linear inverse problems. Whereas regularization-based approach...