{"ID":2852283,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.18654","arxiv_id":"2510.18654","title":"Differentially Private E-Values","abstract":"E-values have gained prominence as flexible tools for statistical inference and risk control, enabling anytime- and post-hoc-valid procedures under minimal assumptions. However, many real-world applications fundamentally rely on sensitive data, which can be leaked through e-values. To ensure their safe release, we propose a general framework to transform non-private e-values into differentially private ones. Towards this end, we develop a novel biased multiplicative noise mechanism that ensures our e-values remain statistically valid. We show that our differentially private e-values attain strong statistical power, and are asymptotically as powerful as their non-private counterparts. Experiments across online risk monitoring, private healthcare, and conformal e-prediction demonstrate our approach's effectiveness and illustrate its broad applicability.","short_abstract":"E-values have gained prominence as flexible tools for statistical inference and risk control, enabling anytime- and post-hoc-valid procedures under minimal assumptions. However, many real-world applications fundamentally rely on sensitive data, which can be leaked through e-values. To ensure their safe release, we prop...","url_abs":"https://arxiv.org/abs/2510.18654","url_pdf":"https://arxiv.org/pdf/2510.18654v1","authors":"[\"Daniel Csillag\",\"Diego Mesquita\"]","published":"2025-10-21T14:03:35Z","proceeding":"stat.ME","tasks":"[\"stat.ME\",\"cs.CR\",\"cs.LG\",\"stat.ML\"]","methods":"[]","has_code":false}