Posterior Uncertainty

Posterior uncertainty quantification aims to accurately represent the confidence in model predictions, crucial for reliable decision-making in various fields. Current research focuses on improving the calibration and efficiency of Bayesian inference methods, employing techniques like diffusion-based models, variational inference with novel regularizations, and physics-informed Bayesian optimization, often within the context of neural networks and other complex models. These advancements address limitations of existing approaches, such as overconfidence or computational cost, leading to more robust and reliable uncertainty estimates in applications ranging from scientific modeling to safety-critical systems. The resulting improvements in uncertainty quantification enhance the trustworthiness and interpretability of complex models across diverse scientific disciplines.

Papers

October 18, 2024

Predictive variational inference: Learn the predictively optimal posterior distribution
Jinlin Lai, Yuling Yao
Variational Inference Posterior Inference Marginal Distribution Posterior Uncertainty Variational Density

July 11, 2024

Inflationary Flows: Calibrated Bayesian Inference with Diffusion-Based Models
Daniela de Albuquerque, John Pearson
Bayesian Inference Posterior Inference Variational Method Posterior Uncertainty

June 10, 2024

Physics-Informed Bayesian Optimization of Variational Quantum Circuits
Kim A. Nicoli, Christopher J. Anders, Lena Funcke, Tobias Hartung, Karl Jansen, Stefan Kühn, Klaus-Robert Müller, Paolo Stornati, Pan Kessel, Shinichi Nakajima
Bayesian Optimization Quantum Circuit Variational Quantum Circuit Physic Based Variational Quantum Quantum Classical Posterior Uncertainty

June 6, 2024

Regularized KL-Divergence for Well-Defined Function-Space Variational Inference in Bayesian neural networks
Tristan Cinquin, Robert Bamler
Variational Inference Bayesian Neural Network KL Divergence Informative Prior Posterior Uncertainty Function Space Variational Inference

March 18, 2024

Posterior Uncertainty Quantification in Neural Networks using Data Augmentation
Luhuan Wu, Sinead Williamson
Neural Network Deep Learning Data Augmentation Uncertainty Quantification Deep Ensemble Posterior Inference Predictive Distribution Posterior Uncertainty

November 14, 2023

Mean-field variational inference with the TAP free energy: Geometric and statistical properties in linear models
Michael Celentano, Zhou Fan, Licong Lin, Song Mei
Variational Inference Mean Field Posterior Inference Free Energy Linear Model Statistical Property Bayesian Regression Posterior Uncertainty

October 21, 2022

Efficient identification of informative features in simulation-based inference
Jonas Beck, Michael Deistler, Yves Bernaerts, Jakob Macke, Philipp Berens
Simulation Based Inference Identification Accuracy Neural Likelihood Surrogate Likelihood Posterior Uncertainty Hodgkin Huxley

April 29, 2022

Tractable Uncertainty for Structure Learning
Benjie Wang, Matthew Wicker, Marta Kwiatkowska
Structure Learning Bayesian Structure Posterior Uncertainty

March 6, 2022

Scalable Uncertainty Quantification for Deep Operator Networks using Randomized Priors
Yibo Yang, Georgios Kissas, Paris Perdikaris
Uncertainty Quantification Deep Operator Network Reliable Uncertainty Model Based Prior Posterior Uncertainty

November 25, 2021

Differentially private stochastic expectation propagation (DP-SEP)
Margarita Vinaroz, Mijung Park
Private Stochastic Gradient Descent Expectation Propagation Posterior Uncertainty