Statistical Analysis of using the Shapley Value for Sensor Anomaly Localization with Accurate Classifiers

2026-05-30

AI Summary

This study investigates the use of Shapley values for anomaly or attack localization in sensor networks and provides a systematic evaluation of their performance under statistically dependent observations. Leveraging optimal binary classifiers and integrating information-theoretic and statistical decision-theoretic principles, the work presents the first rigorous analysis of the statistical performance of Shapley values in anomaly localization. Theoretically, it is shown that under independent observations, the Shapley-based test is equivalent to a simplified single-term test; however, under correlated observations—such as those modeled by bivariate Gaussian or Laplacian distributions—the two approaches exhibit markedly different performance. The paper further introduces a fusion strategy that provably outperforms either method alone. Numerical experiments corroborate the theoretical findings, highlighting both the potential and limitations of Shapley values in settings with statistical dependencies.

Abstract

Recent publications have suggested using the Shap- ley value for sensor anomaly/attack localization. We study the performance of such an approach by using mathematically de- fined optimum binary classifiers in the Shapley value calculation. To judge localization performance, we study the ability of the Shapley value of a given sensor observation to determine if that observation is anomalous. First, we prove that for cases with independent sensor observations, an optimized anomaly test using the Shapley value is equivalent to an optimized lower-complexity anomaly test using a single term in the Shapley value calculation, yielding the exact same probability of error. For some popular dependent observation cases involving two sensors, including correlated bivariate Gaussian/Laplacian probability density functions and constant/Gaussian at- tacks/anomalies, we prove that these two tests are fundamentally different, yielding different decision regions and error probabil- ities. Further, we prove that the Shapley value test is sometimes strictly inferior to the other (single term in Shapley calculation) test in certain statistically dependent bivariate Gaussian scenarios with large correlation magnitude and additive attacks/anomalies, while it is strictly superior in others, depending on the sign of the correlation. One can combine these two approaches to obtain a strictly better approach in these cases. These results, which provide the first theoretical statistical analysis of Shapley-based localization, seem very interesting based on the wide acceptance of the Shapley value by many researchers and should encourage further research on this topic. Numerical results are provided which illustrate our findings.

Problem

• Shapley value
• sensor anomaly localization
• statistical dependence
• optimal binary classifiers
• error probability

Innovation

• Shapley value
• anomaly localization
• optimal binary classifier
• statistical dependence
• error probability