We have 3 of 4 papers
We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.
Not the right person? Other researchers publish under this name.
This work proposes a principled, attack-agnostic approach that uses mutual information to measure what the final model reveals about a datapoint's training membership and shows that this final-model disclosure is upper bounded by the sum of per-iteration gradient disclosures.
The Batch-Space Ghost (BS-Ghost) algorithm makes these scores practical inside the training loop through shared computation in batch space, without storing model-sized example gradients, and demonstrates that BGU identifies data that causally shapes final behavior.
This work introduces minimal information disclosure (MID), which designs and quantifies the information content of verifier-facing evidence itself and measures collateral leakage with conditional mutual information.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
