<ul data-eligibleForWebStory="true"><li>Researchers introduce WaKA (Wasserstein K-nearest-neighbors Attribution), an attribution method that combines principles from LiRA and k-nearest neighbors classifiers.</li><li>WaKA measures the contribution of individual data points to a model's loss distribution without needing to sample subsets of the training set.</li><li>It can be used as a membership inference attack (MIA) to assess privacy risks or for privacy influence measurement and data valuation.</li><li>WaKA bridges the gap between data attribution and MIA by distinguishing a data point's value from its privacy risk.</li><li>Self-attribution values in WaKA have a stronger correlation with attack success rates than a point's contribution to model generalization.</li><li>WaKA performs closely to LiRA in MIA tasks on k-NN classifiers but with better computational efficiency.</li><li>It demonstrates greater robustness than Shapley Values for data minimization tasks on imbalanced datasets.</li></ul>

WaKA: Data Attribution using K-Nearest Neighbors and Membership Privacy Principles

Discover more