Epistemic Uncertainty Quantification

Epistemic uncertainty quantification (EUQ) aims to measure a model's confidence in its predictions, distinguishing it from aleatoric uncertainty (inherent randomness in the data). Current research focuses on improving EUQ methods for various model architectures, including Bayesian neural networks and pre-trained models, often employing gradient-based approaches or techniques like evidential deep learning to estimate uncertainty. Accurate EUQ is crucial for building trustworthy AI systems, particularly in high-stakes applications like healthcare and autonomous driving, by enabling better decision-making and improved out-of-distribution detection.

Papers

October 7, 2024

PH-Dropout: Prctical Epistemic Uncertainty Quantification for View Synthesis
Chuanhao Sun, Thanos Triantafyllou, Anthos Makris, Maja Drmač, Kai Xu, Luo Mai, Mahesh K. Marina
Neural Radiance Field Novel View Synthesis View Synthesis Epistemic Uncertainty Epistemic Uncertainty Quantification

July 2, 2024

The Epistemic Uncertainty Hole: an issue of Bayesian Neural Networks
Mohammed Fellaji, Frédéric Pennerath
Bayesian Neural Network Epistemic Uncertainty Bayesian Deep Learning Proper Issue Labeling Epistemic Uncertainty Quantification

June 4, 2024

April 15, 2024

Epistemic Uncertainty Quantification For Pre-trained Neural Network
Hanjing Wang, Qiang Ji
Epistemic Uncertainty Uncertainty Calibration Epistemic Uncertainty Quantification

February 20, 2024

Investigating the Impact of Model Instability on Explanations and Uncertainty
Sara Vera Marjanović, Isabelle Augenstein, Christina Lioma
Global Impact Line by Line Explanation High Uncertainty Anticipation Epistemic Uncertainty Explanation Quality Epistemic Uncertainty Quantification Model Stability

February 14, 2024

Is Epistemic Uncertainty Faithfully Represented by Evidential Deep Learning Methods?
Mira Jürgens, Nis Meinert, Viktor Bengs, Eyke Hüllermeier, Willem Waegeman
Loss Function Epistemic Uncertainty Evidential Deep Learning Epistemic Uncertainty Quantification

February 13, 2024

Experts Don't Cheat: Learning What You Don't Know By Predicting Pairs
Daniel D. Johnson, Daniel Tarlow, David Duvenaud, Chris J. Maddison
Generative Model Expert Knowledge Epistemic Uncertainty Aleatoric Uncertainty Epistemic Uncertainty Quantification Pair Selection

August 15, 2023

Benchmarking Scalable Epistemic Uncertainty Quantification in Organ Segmentation
Jadie Adams, Shireen Y. Elhabian
Epistemic Uncertainty Accurate Segmentation Organ Segmentation Epistemic Uncertainty Quantification

January 30, 2023

On Second-Order Scoring Rules for Epistemic Uncertainty Quantification
Viktor Bengs, Eyke Hüllermeier, Willem Waegeman
Empirical Risk Minimization Second Order Language Learner Scoring Rule Epistemic Uncertainty Quantification

November 16, 2022

Interpretable Self-Aware Neural Networks for Robust Trajectory Prediction
Masha Itkina, Mykel J. Kochenderfer
Autonomous Driving Autonomous Vehicle Trajectory Prediction Evidential Deep Learning Epistemic Uncertainty Quantification

November 11, 2022

The Implicit Delta Method
Nathan Kallus, James McInerney
Uncertainty Quantification Epistemic Uncertainty Quantification Delta Method Computational Uncertainty

August 22, 2022

Some Supervision Required: Incorporating Oracle Policies in Reinforcement Learning via Epistemic Uncertainty Metrics
Jun Jet Tai, Jordan K. Terry, Mauro S. Innocente, James Brusey, Nadjim Horri
Reinforcement Learning Reinforcement Learning Benchmark Sparse Reward Environment Test Oracle Actor Critic Reinforcement Learning Epistemic Uncertainty Quantification Uncertainty Aware Exploration

July 22, 2022

Complementing Semi-Supervised Learning with Uncertainty Quantification
Ehsan Kazemi
Semi Supervised Learning Semi Supervised Uncertainty Quantification Pseudo Label Adaptive Loss Epistemic Uncertainty Quantification

March 11, 2022

Pitfalls of Epistemic Uncertainty Quantification through Loss Minimisation
Viktor Bengs, Eyke Hüllermeier, Willem Waegeman
Loss Function Uncertainty Quantification Second Order Epistemic Uncertainty Common Pitfall Loss Minimization Language Learner Epistemic Uncertainty Quantification

November 18, 2021

Exploring the Limits of Epistemic Uncertainty Quantification in Low-Shot Settings
Matias Valdenegro-Toro
Uncertainty Quantification Distribution Detection AI System Continuum Limit Epistemic Uncertainty Uncertainty Estimation Method Epistemic Uncertainty Quantification