ppAURORA: Privacy Preserving Area Under Receiver Operating Characteristic and Precision-Recall Curves

TL;DR

ppAURORA is a novel privacy-preserving multi-party computation method that accurately computes the AUC and PR curves for sensitive data without revealing individual data points, enabling secure model evaluation.

Contribution

It introduces the first exact privacy-preserving solution for computing AUC and PR curves using MPC, handling ties and ensuring data privacy in collaborative settings.

Findings

Accurately computes AUC and PR curves on sensitive data.

Maintains privacy while matching plaintext evaluation results.

Demonstrates scalability with synthetic data experiments.

Abstract

Computing an AUC as a performance measure to compare the quality of different machine learning models is one of the final steps of many research projects. Many of these methods are trained on privacy-sensitive data and there are several different approaches like $\epsilon$-differential privacy, federated machine learning and cryptography if the datasets cannot be shared or used jointly at one place for training and/or testing. In this setting, it can also be a problem to compute the global AUC, since the labels might also contain privacy-sensitive information. There have been approaches based on $\epsilon$-differential privacy to address this problem, but to the best of our knowledge, no exact privacy preserving solution has been introduced. In this paper, we propose an MPC-based solution, called ppAURORA, with private merging of individually sorted lists from multiple sources to compute the exact AUC as one could obtain on the pooled original test samples. With ppAURORA, the computation of the exact area under precision-recall and receiver operating characteristic curves is possible even when ties between prediction confidence values exist. We use ppAURORA to evaluate two different models predicting acute myeloid leukemia therapy response and heart disease, respectively. We also assess its scalability via synthetic data experiments. All these experiments show that we efficiently and privately compute the exact same AUC with both evaluation metrics as one can obtain on the pooled test samples in plaintext according to the semi-honest adversary setting.