Paper

Mind the Gap: Mixtures of Gaussians in Approximate Differential Privacy

Unreadunread

AI summary

arXiv:2605. 28078v1 Announce Type: new Abstract: We design a class of additive noise mechanisms that satisfy ((\varepsilon, \delta))-differential privacy (DP) for scalar, real-valued query functions with known sensitivities, with a particular focus on moderate and low-privacy regimes.

Reading status

Priority

About this paper

Authors: Huikang Liu, Aras Selvi, Wolfram Wiesemann
Released: 2026-05-28 (2mo ago)
Topic: Related to current project
Relation to my work: Read next because Mind the Gap: Mixtures of Gaussians in Approximate Differential Privacy overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, rate, compare. Source: arxiv cs.CR (Cryptography and Security).
arXiv: 2605.28078

Ask Claude about this paper

Posts a comment with @claude so the runner spins up a Claude Code session scoped to Mind the Gap: Mixtures of Gaussians in Approximate Differential Privacy. Replies appear in the conversation below.

Comments

No comments yet.

Related records

No links yet.