diff options
author | Yuchen Pei <me@ypei.me> | 2019-03-14 10:02:09 +0100 |
---|---|---|
committer | Yuchen Pei <me@ypei.me> | 2019-03-14 10:02:09 +0100 |
commit | 48b6240950859e6123e717fae704628c5a1c0fdb (patch) | |
tree | c8ffb13e9827c7d773c5d81d9860538c90946242 | |
parent | d667a457985edf012759e9fd12b66807583dcc1e (diff) |
minor
-rw-r--r-- | posts/2019-03-13-a-tail-of-two-densities.md | 4 |
1 files changed, 4 insertions, 0 deletions
diff --git a/posts/2019-03-13-a-tail-of-two-densities.md b/posts/2019-03-13-a-tail-of-two-densities.md index fbf3127..f7740e2 100644 --- a/posts/2019-03-13-a-tail-of-two-densities.md +++ b/posts/2019-03-13-a-tail-of-two-densities.md @@ -918,6 +918,10 @@ when $\epsilon$ is sufficiently small, in which case the leading term is of orde $O(\sqrt k \epsilon)$ and we save a $\sqrt k$ in the $\epsilon$-part compared to the Basic Composition Theorem (Claim 10). +**Remark**. In practice one can try different choices of $\beta$ and settle +with the one that gives the best privacy guarantee. +See the discussions at the end of Part 2 of this post. + **Proof**. Let $p_i$, $q_i$, $p$ and $q$ be the laws of $M_i(x)$, $M_i(x')$, $M(x)$ and $M(x')$ respectively. |