Just rotate it! Uncertainty estimation in closed-source models via multiple queries

TL;DR

This paper introduces a simple method for estimating uncertainty in closed-source image classification models by applying natural transformations like rotations to create multiple queries, significantly improving calibration over naive confidence assignments.

Contribution

The paper demonstrates that natural transformations outperform Gaussian noise for uncertainty estimation and introduces a transfer learning approach to further enhance calibration.

Findings

Natural transformations improve uncertainty calibration.

Empirical results show better calibration with transformations.

Theoretical analysis explains why natural transformations outperform Gaussian noise.

Abstract

We propose a simple and effective method to estimate the uncertainty of closed-source deep neural network image classification models. Given a base image, our method creates multiple transformed versions and uses them to query the top-1 prediction of the closed-source model. We demonstrate significant improvements in the calibration of uncertainty estimates compared to the naive baseline of assigning 100\% confidence to all predictions. While we initially explore Gaussian perturbations, our empirical findings indicate that natural transformations, such as rotations and elastic deformations, yield even better-calibrated predictions. Furthermore, through empirical results and a straightforward theoretical analysis, we elucidate the reasons behind the superior performance of natural transformations over Gaussian noise. Leveraging these insights, we propose a transfer learning approach that further improves our calibration results.