Deep Classifiers with Label Noise Modeling and Distance Awareness

TL;DR

This paper introduces HetSNGP, a method that jointly models model and data uncertainty in deep classifiers, improving out-of-distribution detection and calibration, especially when combined with ensemble techniques.

Contribution

The paper presents HetSNGP, a novel approach that combines distance-aware and label uncertainty modeling, and extends it with HetSNGP Ensemble for enhanced uncertainty estimation.

Findings

Outperforms baseline methods on CIFAR-100C, ImageNet-C, and ImageNet-A datasets.

Effectively combines model and data uncertainty for better reliability.

Ensemble version further improves uncertainty estimation and out-of-distribution detection.

Abstract

Uncertainty estimation in deep learning has recently emerged as a crucial area of interest to advance reliability and robustness in safety-critical applications. While there have been many proposed methods that either focus on distance-aware model uncertainties for out-of-distribution detection or on input-dependent label uncertainties for in-distribution calibration, both of these types of uncertainty are often necessary. In this work, we propose the HetSNGP method for jointly modeling the model and data uncertainty. We show that our proposed model affords a favorable combination between these two types of uncertainty and thus outperforms the baseline methods on some challenging out-of-distribution datasets, including CIFAR-100C, ImageNet-C, and ImageNet-A. Moreover, we propose HetSNGP Ensemble, an ensembled version of our method which additionally models uncertainty over the network parameters and outperforms other ensemble baselines.