Being Bayesian, Even Just a Bit, Fixes Overconfidence in ReLU Networks

Being Bayesian, Even Just a Bit, Fixes Overconfidence in ReLU Networks

24 February 2020

Agustinus Kristiadi

Matthias Hein

Papers citing "Being Bayesian, Even Just a Bit, Fixes Overconfidence in ReLU Networks"

5 / 55 papers shown

Title
Know Your Limits: Uncertainty Estimation with ReLU Classifiers Fails at Reliable OOD Detection Dennis Ulmer Giovanni Cina OODD 27 31 0 09 Dec 2020
Deep Learning is Singular, and That's Good Daniel Murfet Susan Wei Mingming Gong Hui Li Jesse Gell-Redman T. Quella UQCV 16 26 0 22 Oct 2020
Estimating Model Uncertainty of Neural Networks in Sparse Information Form Jongseo Lee Matthias Humt Jianxiang Feng Rudolph Triebel BDL UQCV 25 46 0 20 Jun 2020
Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles Balaji Lakshminarayanan Alexander Pritzel Charles Blundell UQCV BDL 270 5,660 0 05 Dec 2016
Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning Y. Gal Zoubin Ghahramani UQCV BDL 279 9,136 0 06 Jun 2015