Decomposition of Uncertainty in Bayesian Deep Learning for Efficient and Risk-sensitive Learning

By Stefan Depeweg, José Miguel Hernández-Lobato, Finale Doshi-Velez, Steffen Udluft. International Conference on Machine Learning, 2017

521 citations recorded by the source corpus.

Subfields: Probabilistic & Bayesian Methods

View Paper: https://www.semanticscholar.org/paper/a9c8124b18c0e843878f2df4d49da6f250dc06e7

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Probabilistic & Bayesian Methods

Citations in the corpus

Getting a CLUE: A Method for Explaining Uncertainty Estimates (2020, methodology): “It enables the former to be used for uncertainty aware tasks, such as OOD detection (Daxberger & Hernández-Lobato, 2019), continual learning (Nguyen et al., 2018), active learning (Depeweg et al., 2018), and Bayesian optimization (Springenberg et al., 2016).”
Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models (2018, background): “An advantage of using TS∞ is that aleatoric and epistemic uncertainties are separable [Depeweg et al., 2018].”
Hybrid Models with Deep and Invertible Features (2019, background): “, 2014), information regularization (Szummer & Jaakkola, 2003), and decomposition of uncertainty (Depeweg et al., 2018).”
Overcoming Model Bias for Robust Offline Deep Reinforcement Learning (2020, methodology): “This advantage over model-free reinforcement learning has been demonstrated many times over (Deisenroth and Rasmussen 2011; Depeweg et al. 2017; Kurutach et al. 2018; Nagabandi et al. 2018); however, it usually comes with the downside of lower asymptotic performance due to model bias (Nagabandi et a”
Aleatoric and epistemic uncertainty in machine learning: an introduction to concepts and methods (2019, background): “An explicit attempt at measuring and separating aleatoric and epistemic uncertainty (in the context of regression) is made by Depeweg et al. (2018). Their idea is as follows: They quantify the total uncertainty as well as the aleatoric uncertainty, and then obtain the epistemic uncertainty as the di”
A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges (2020, background): “[229] exhibited the decomposition and derived uncertainty into aleatoric and epistemic for decision support systems.”
Hands-On Bayesian Neural Networks—A Tutorial for Deep Learning Users (2020, background): “lack of knowledge, measured by p(θ|D), which can be reduced with more data, and aleatoricuncertainty, i.e., the uncertainty due to the (partially) aleatoric nature of the data and measured by p(y|x,θ)[14, 44]. This makes BNNs very data efficient, as they can learn from a small dataset without overfi”
A Review and Comparative Study on Probabilistic Object Detection in Autonomous Driving (2020, methodology): “Multiple techniques for generating approximate solutions of the posterior distribution in Equation (1) have been proposed in the literature, including Variational Inference (VI) [24]–[26], Markov Chain Monte Carlo (MCMC) [27]– [29], Stochastic Gradient Descent (SGD) Approximation [30]– [32], and Lap”
Estimating Uncertainty and Interpretability in Deep Learning for Coronavirus (COVID-19) Detection (2020, background): “However predictive uncertainty in deep learning actually results from two separate forms of uncertainty [6]:”
Understanding Uncertainty in Bayesian Neural Networks (2019, methodology): “Bayesian methods enable NNs to be used for tasks which require model uncertainty, such as continual learning (Nguyen et al., 2018), active learning (Depeweg et al., 2018), and Bayesian optimisation (Springenberg et al., 2016).”
Single-Model Uncertainties for Deep Learning (2018, background): “Combined uncertainty While the idea of a combined measure of uncertainty is appealing, aleatoric and epistemic unknowns measure different quantities in different units; different applications will highlight the importance of one over the other [19, 44].”
Sequential Neural Methods for Likelihood-free Inference (2018, background): “Recent work [2, 4, 7] has highlighted the concepts of aleatoric (referring to stochasticity inherent to our data) and epistemic uncertainty (traditional model uncertainty), discussing methods to disentangle these two confounding factors in deep learning.”
A survey of uncertainty in deep neural networks (2021, unknown): “of aleatoric uncertainty [139], [140], [141].”