Bias-Reduced Uncertainty Estimation for Deep Neural Classifiers

Abstract

We consider the problem of uncertainty estimation in the context of (non-Bayesian) deep neural classification. In this context, all known methods are based on extracting uncertainty signals from a trained network optimized to solve the classification problem at hand. We demonstrate that such techniques tend to introduce biased estimates for instances whose predictions are supposed to be highly confident. We argue that this deficiency is an artifact of the dynamics of training with SGD-like optimizers, and it has some properties similar to overfitting. Based on this observation, we develop an uncertainty estimation algorithm that selectively estimates the uncertainty of highly confident points, using earlier snapshots of the trained model, before their estimates are jittered (and way before they are ready for actual classification). We present extensive experiments indicating that the proposed algorithm provides uncertainty estimates that are consistently better than all known methods.

arXiv commentAccepted to ICLR 2019

Cited by in corpus (13)

• WILDS: A Benchmark of in-the-Wild Distribution Shifts
• Confidence Calibration for Convolutional Neural Networks Using Structured Dropout
• On Last-Layer Algorithms for Classification: Decoupling Representation from Uncertainty Estimation
• Unified Classification and Rejection: A One-versus-All Framework
• Dense Uncertainty Estimation
• A Unified Benchmark for the Unknown Detection Capability of Deep Neural Networks
• DOCTOR: A Simple Method for Detecting Misclassification Errors
• Controlling Risk of Web Question Answering
• Classification Confidence Estimation with Test-Time Data-Augmentation
• Selective Classification Can Magnify Disparities Across Groups
• Graceful Degradation and Related Fields
• Reinforcement Learning Under Algorithmic Triage
• On Deep Neural Network Calibration by Regularization and its Impact on Refinement