Deep Classifiers with Label Noise Modeling and Distance Awareness

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.

arXiv commentPublished in TMLR

References in corpus (29)

• Scikit-learn: Machine Learning in Python
• An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
• What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?
• Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift
• Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles
• Evidential Deep Learning to Quantify Classification Uncertainty
• Predictive Uncertainty Estimation via Prior Networks
• High-Quality Prediction Intervals for Deep Learning: A Distribution-Free, Ensembled Approach
• Decomposition of Uncertainty in Bayesian Deep Learning for Efficient and Risk-sensitive Learning
• Simple and Principled Uncertainty Estimation with Deterministic Deep Learning via Distance Awareness
• Stochastic Segmentation Networks: Modelling Spatially Correlated Aleatoric Uncertainty
• Are we done with ImageNet?
• Single-Model Uncertainties for Deep Learning
• A Systematic Comparison of Bayesian Deep Learning Robustness in Diabetic Retinopathy Tasks
• Posterior Network: Uncertainty Estimation without OOD Samples via Density-Based Pseudo-Counts
• Noise Contrastive Priors for Functional Uncertainty
• Simple, Distributed, and Accelerated Probabilistic Programming
• Hydra: Preserving Ensemble Diversity for Model Distillation
• Dirichlet-based Gaussian Processes for Large-scale Calibrated Classification
• Training independent subnetworks for robust prediction
• Bayesian Neural Network Priors Revisited
• BNNpriors: A library for Bayesian neural network inference with different prior distributions
• Repulsive Deep Ensembles are Bayesian
• A Simple Probabilistic Method for Deep Classification under Input-Dependent Label Noise
• Cold Posteriors and Aleatoric Uncertainty
• Output-Constrained Bayesian Neural Networks
• Fast Predictive Uncertainty for Classification with Bayesian Deep Networks
• Variational Variance: Simple, Reliable, Calibrated Heteroscedastic Noise Variance Parameterization
• On Stein Variational Neural Network Ensembles