Credal Bayesian

Credal Bayesian methods extend traditional Bayesian approaches by representing uncertainty not as a single probability distribution, but as a set of possible distributions (a "credal set"). This framework is particularly valuable for handling noisy data, model uncertainty, and out-of-distribution predictions, with current research focusing on applications within deep learning, including Bayesian neural networks and deep ensembles, often incorporating techniques like robust loss functions and probabilistic circuits for improved uncertainty quantification. The resulting enhanced robustness and improved uncertainty estimation have significant implications for various fields, such as medical image analysis and autonomous systems, where reliable predictions under uncertainty are crucial.

Papers

September 25, 2024

SSP-RACL: Classification of Noisy Fundus Images with Self-Supervised Pretraining and Robust Adaptive Credal Loss
Mengwen Ye, Yingzi Huangfu, You Li, Zekuan Yu
Classification Code Masked Autoencoders Fundus Image Self Supervised Pretraining Credal Bayesian

May 23, 2024

Credal Wrapper of Model Averaging for Uncertainty Estimation on Out-Of-Distribution Detection
Kaizheng Wang, Fabio Cuzzolin, Keivan Shariatmadar, David Moens, Hans Hallez
Distribution Detection Uncertainty Estimation Bayesian Neural Network Deep Ensemble Neural Network Baseline Model Averaging Credal Set New Wrapper Method Credal Bayesian

February 1, 2024

Credal Learning Theory
Michele Caprio, Maryam Sultana, Eleni Elia, Fabio Cuzzolin
Machine Learning Convex Set Statistical Learning Theory Hypothesis Space Credal Set Credal Bayesian

February 27, 2023

Markov Conditions and Factorization in Logical Credal Networks
Fabio Gagliardi Cozman
Prime Factorization Credal Bayesian

February 19, 2023

Credal Bayesian Deep Learning
Michele Caprio, Souradeep Dutta, Kuk Jin Jang, Vivian Lin, Radoslav Ivanov, Oleg Sokolsky, Insup Lee
Uncertainty Quantification Bayesian Neural Network Bayesian Model Averaging Credal Bayesian

August 4, 2022

Credal Valuation Networks for Machine Reasoning Under Uncertainty
Branko Ristic, Alessio Benavoli, Sanjeev Arulampalam
High Uncertainty Anticipation Automated Reasoning Valuation Model Credal Bayesian Imprecise Probability

May 11, 2022

Robustness Guarantees for Credal Bayesian Networks via Constraint Relaxation over Probabilistic Circuits
Hjalmar Wijk, Benjie Wang, Marta Kwiatkowska
Robustness Guarantee Probabilistic Circuit Bayesian Network Structure Constraint Relaxation Credal Bayesian