PAC Confidence Predictions for Deep Neural Network Classifiers

Sangdon Park, Shuo Li, Insup Lee, Osbert Bastani

Keywords: classification, calibration, probably approximated correct guarantee, fast DNN inference, safe planning

Abstract: A key challenge for deploying deep neural networks (DNNs) in safety critical settings is the need to provide rigorous ways to quantify their uncertainty. In this paper, we propose a novel algorithm for constructing predicted classification confidences for DNNs that comes with provable correctness guarantees. Our approach uses Clopper-Pearson confidence intervals for the Binomial distribution in conjunction with the histogram binning approach to calibrated prediction. In addition, we demonstrate how our predicted confidences can be used to enable downstream guarantees in two settings: (i) fast DNN inference, where we demonstrate how to compose a fast but inaccurate DNN with an accurate but slow DNN in a rigorous way to improve performance without sacrificing accuracy, and (ii) safe planning, where we guarantee safety when using a DNN to predict whether a given action is safe based on visual observations. In our experiments, we demonstrate that our approach can be used to provide guarantees for state-of-the-art DNNs.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

One-sentence Summary: We propose a novel algorithm for constructing predicted classification confidences for DNNs that comes with provable correctness guarantees, and demonstrate how our predicted confidences can be used to enable downstream guarantees in two settings.