Posterior Network: Uncertainty Estimation without OOD Samples via Density-Based Pseudo-Counts

TL;DR

The Posterior Network (PostNet) employs Normalizing Flows to estimate uncertainty in predictions without needing out-of-distribution samples during training, achieving state-of-the-art results in OOD detection and calibration.

Contribution

PostNet introduces a novel density-based approach using Normalizing Flows to model posterior distributions for uncertainty estimation without OOD data during training.

Findings

PostNet accurately reflects uncertainty for in- and out-of-distribution data.

Achieves state-of-the-art OOD detection performance.

Improves uncertainty calibration under dataset shifts.

Abstract

Accurate estimation of aleatoric and epistemic uncertainty is crucial to build safe and reliable systems. Traditional approaches, such as dropout and ensemble methods, estimate uncertainty by sampling probability predictions from different submodels, which leads to slow uncertainty estimation at inference time. Recent works address this drawback by directly predicting parameters of prior distributions over the probability predictions with a neural network. While this approach has demonstrated accurate uncertainty estimation, it requires defining arbitrary target parameters for in-distribution data and makes the unrealistic assumption that out-of-distribution (OOD) data is known at training time. In this work we propose the Posterior Network (PostNet), which uses Normalizing Flows to predict an individual closed-form posterior distribution over predicted probabilites for any input sample. The posterior distributions learned by PostNet accurately reflect uncertainty for in- and out-of-distribution data -- without requiring access to OOD data at training time. PostNet achieves state-of-the art results in OOD detection and in uncertainty calibration under dataset shifts.