Bayesian neural networks with latent variables are scalable and flexible
probabilistic models: They account for uncertainty in the estimation of the
network weights and, by making use of latent variables, can capture complex
noise patterns in the data. We show how to extract and decompose uncertainty
into epistemic and aleatoric components for decision-making purposes. This
allows us to successfully identify informative points for active learning of
functions with heteroscedastic and bimodal noise. Using the decomposition we
further define a novel risk-sensitive criterion for reinforcement learning to
identify policies that balance expected cost, model-bias and noise aversion.
Description
[1710.07283] Decomposition of Uncertainty in Bayesian Deep Learning for Efficient and Risk-sensitive Learning
%0 Journal Article
%1 depeweg2017decomposition
%A Depeweg, Stefan
%A Hernández-Lobato, José Miguel
%A Doshi-Velez, Finale
%A Udluft, Steffen
%D 2017
%K bayesian uncertainty
%T Decomposition of Uncertainty in Bayesian Deep Learning for Efficient and
Risk-sensitive Learning
%U http://arxiv.org/abs/1710.07283
%X Bayesian neural networks with latent variables are scalable and flexible
probabilistic models: They account for uncertainty in the estimation of the
network weights and, by making use of latent variables, can capture complex
noise patterns in the data. We show how to extract and decompose uncertainty
into epistemic and aleatoric components for decision-making purposes. This
allows us to successfully identify informative points for active learning of
functions with heteroscedastic and bimodal noise. Using the decomposition we
further define a novel risk-sensitive criterion for reinforcement learning to
identify policies that balance expected cost, model-bias and noise aversion.
@article{depeweg2017decomposition,
abstract = {Bayesian neural networks with latent variables are scalable and flexible
probabilistic models: They account for uncertainty in the estimation of the
network weights and, by making use of latent variables, can capture complex
noise patterns in the data. We show how to extract and decompose uncertainty
into epistemic and aleatoric components for decision-making purposes. This
allows us to successfully identify informative points for active learning of
functions with heteroscedastic and bimodal noise. Using the decomposition we
further define a novel risk-sensitive criterion for reinforcement learning to
identify policies that balance expected cost, model-bias and noise aversion.},
added-at = {2019-11-25T01:19:12.000+0100},
author = {Depeweg, Stefan and Hernández-Lobato, José Miguel and Doshi-Velez, Finale and Udluft, Steffen},
biburl = {https://www.bibsonomy.org/bibtex/2dedd46eab9ce60e6c9d1983a1541d7c3/kirk86},
description = {[1710.07283] Decomposition of Uncertainty in Bayesian Deep Learning for Efficient and Risk-sensitive Learning},
interhash = {440e49ace6bf0ab5eb7493237f047ada},
intrahash = {dedd46eab9ce60e6c9d1983a1541d7c3},
keywords = {bayesian uncertainty},
note = {cite arxiv:1710.07283Comment: This paper supersedes arXiv:1706.08495},
timestamp = {2019-11-25T01:19:12.000+0100},
title = {Decomposition of Uncertainty in Bayesian Deep Learning for Efficient and
Risk-sensitive Learning},
url = {http://arxiv.org/abs/1710.07283},
year = 2017
}