CS Paper Radar

Tight Bounds for Differentially Private Cut Release

Chenghua Liu, Boning Meng

cs.DS ·submitted 2026-11-14 18:00 UTC · v1
theory

Abstract

We give a polynomial-time algorithm for releasing a nonnegative weighted synthetic graph under $(\varepsilon,δ)$-edge differential privacy. Given an $n$-vertex nonnegative weighted graph $G=(V,w_G)$, the output $H=(V,w_H)$ satisfies, with probability at least $1-β$, simultaneously for every $S\subseteq V$, $|w_G(S,V\setminus S)-w_H(S,V\setminus S)| \leγw_G(S,V\setminus S)+O\bigl((n\log n)/(\varepsilonγ)\bigr)$, provided that $γ,\varepsilon,δ,β$ are at least inverse polynomial in $n$. For constant $γ,\varepsilon$ and $δ=β=n^{-10}$, this improves the recently obtained additive error bound of $O(n\log^7 n)$ to $O(n\log n)$. We also prove a matching $Ω(n\log n)$ lower bound. Our algorithm combines degree decomposition with private multiplicative weights, using a degree-sensitive semidefinite oracle to generate update proposals from single Gaussian samples. Each proposal is validated by a noisy check of its exact entropy decrease, and a joint Rényi analysis of sampling and validation avoids additional logarithmic losses.

Comment
24 pages

Closest papers

Nearest neighbours by meaning across the whole index.

Polynomial-Time Differentially Private Synthetic Graph Release with Nearly-Linear Additive Error

cs.DS math.ST ·2026-11-04 ·Marek Eliáš, Gautam Kamath, Janardhan Kulkarni, George Z. Li, Jingcheng Liu +5 more
theory similarity 0.93
Abstract

We give a polynomial-time $(\varepsilon, δ)$-edge-differentially private algorithm that releases a synthetic graph approximately preserving all cut values of a nonnegative weighted input graph $G$. Our algorithm outputs a nonnegative weighted graph $H$ such that, with probability at least $1-β$, simultaneously for all $S\subseteq V$, \[ \left|w_G(S,V\setminus S)-w_H(S,V\setminus S)\right| \leq γw_G(S,V\setminus S) + O\left(\frac{n}{γ\varepsilon} \log^7 \frac{n}{δβγ\varepsilon}\right). \] For fixed $γ, \varepsilon$ and $δ= β= n^{-10}$, the additive error is $n \operatorname{polylog} n$, improving upon the previous efficient bound of $n^{13/12 + o(1)}$. Furthermore, we establish an $Ω(n/γ\varepsilon)$ lower bound for the relevant parameter regime, showing that our result is tight up to polylogarithmic factors. Our algorithm uses a log-det-regularized SDP to generate fractional queries for multiplicative-weights updates and certify accuracy for all cuts. The privacy analysis charges privacy loss only fractionally per mechanism call, based on the amount of mass on edge coordinates retired from subsequent iterations. Our synthetic graph release yields improved error bounds for several downstream tasks in private graph analysis including correlation clustering, Max-Cut, Max-$k$-Cut, constrained cut problems and hierarchical clustering. Most notably, for unweighted complete-graph correlation clustering, we achieve the optimal $\widetilde O(n)$ additive error while reducing the best previously known approximation ratio for this additive error from several thousand to $1.387$. We also provide a formalization of our results in Lean.

0.53

Lower Bounds for Private Graph Optimization Problems using Reconstruction Attacks

cs.DS cs.CR ·2026-09-09 ·Jacob Imola, Rasmus Pagh, Lukas Retschmeier
top venuetheory similarity 0.77
Abstract · 32 pages, 3 figures. To appear at PODS 2027

This paper studies fundamental graph optimization problems under differential privacy (DP) and shows new, reconstruction-based lower bounds. We consider a graph $G = (V, E, \vec{w})$ where the vertex set $V$ and edges $E$ are public and the weights $\mathbf{w}:E\rightarrow \mathbb{R}$ must be kept differentially private under an $\ell_1$ neighboring relation. For the problems of releasing a minimum-weight spanning tree and a minimum-weight perfect matching, we show new, tight error bounds of $Ω(n\cdot\log(m/n)/ε)$ on worst-case graphs with $n$ vertices and $m>2n$ edges. The upper bounds are known pure DP algorithms while the new lower bound holds even under approximate $(\varepsilon,δ)$-DP as long as $δ\leq (n/m)^{Ω(1)}$. Our lower bounds improve the $Ω(n/ε)$ lower bounds of Sealfon (PODS~'16). The fact that approximate DP does not reduce error for MST under the $\ell_1$ neighboring relation contrasts with the recent upper bound of Pagh et al. (PODS~'25) which shows that approximate DP allows much better error under the $\ell_\infty$ neighboring relation. Going beyond worst-case graphs, we give lower bounds for large families of sparse graphs with expansion properties. We show a lower bound of $Ω(n / ε)$ for the minimum spanning tree for any graph where the minimum cut is at least $Ω(\log(n))$. Finally, we consider the problem of private hierarchical clustering under Dasgupta's cost function (STOC~'16) and show the first approximate DP lower bound parameterized by the minimum weight of a balanced cut. This extends lower bounds of Deng et al. (ICLR~'25) to general graphs and to approximate DP.

0.59

The Power of Interaction in Locally Edge-Private Subgraph Counting

cs.DS cs.CR ·2026-11-03 ·Xian Chen, Pan Peng
theory similarity 0.69
Abstract · 64 pages

We study (non-induced) subgraph counting under interactive local edge differential privacy (LEDP). Given a fixed pattern graph $H$ with $k$ vertices, $\ell\ge1$ edges, and $q$ connected components, our goal is to approximately count its copies in an arbitrary $n$-vertex undirected graph while preserving edge privacy. We give two complementary accuracy guarantees. First, for every $H$, we give Linear Noise Cancellation (LNC), a two-round $\varepsilon$-LEDP algorithm with additive error and root mean square error both $O(n^{k-3/2}/\varepsilon^\ell)$ on every $n$-vertex graph. For fixed sufficiently small $\varepsilon$, this improves the optimal noninteractive additive error for triangles by a factor of $\sqrt n$ and gives the same improvement in worst-case root mean square error for all cliques on at least three vertices. To our knowledge, this is the first separation between interactive and noninteractive LEDP subgraph counting, showing that a single additional round can improve worst-case accuracy polynomially. Second, we determine how accuracy improves when the graph has maximum degree at most a public bound $d$. Our Predict and Correct (PC) framework gives an $(\ell+2)$-round $\varepsilon$-LEDP algorithm with additive error $O(n^{q-1/2}d^{k-q-1}\log^{3\ell-2}n/\varepsilon^\ell)$. For connected patterns, its polynomial factor is $\sqrt n\,d^{k-2}$, giving sublinear error on constant-degree graphs at fixed privacy. We prove matching lower bounds in $n$ and $d$, up to logarithmic factors at fixed privacy, even under arbitrary interaction and approximate privacy with sufficiently small $δ$. At $d=Θ(n)$, the lower bound also establishes the optimality of LNC's dependence on $n$.

0.21

Smoothed Analysis for Distributed MIS

cs.DC ·2026-11-12 ·Dipan Dey, Fabien Dufoulon, William K. Moses, Gopal Pandurangan
theory similarity 0.69
Abstract

We study smoothed analysis of the distributed Maximal Independent Set (MIS) problem. Luby's algorithm computes an MIS in $O(\log n)$ rounds, while the recent $Ω(\log n/\log\log n)$ lower bound of Khoury and Schild (2026) shows that it is nearly optimal even in the LOCAL model. We ask whether small random perturbations can overcome this worst-case barrier. Given an arbitrary base graph $G$ on $n$ nodes, we consider the smoothed graph $\mathcal{G}_s=G\triangle G(n,p)$, where $G(n,p)$ is an independent random graph and $\triangle$ denotes symmetric difference. Thus, each possible edge is independently toggled with probability $p$. We establish nearly tight bounds on the smoothing needed for faster distributed MIS. For every base graph $G$, when $p=β\ln n/n$ for a sufficiently large constant $β$, a randomized LOCAL algorithm, given only $\mathcal{G}_s$, computes an MIS of $\mathcal{G}_s$ in $O(\sqrt{\log n})$ rounds with high probability. This is almost quadratically faster than the worst-case lower bound. Conversely, if $p\le(1/2-δ)\ln n/n$ for a fixed constant $δ>0$, there exists a base graph on which every randomized LOCAL MIS algorithm requires $Ω(\log n/\log\log n)$ rounds on the smoothed graph. Hence, the smoothing parameter required for this speedup is optimal up to a constant factor. These results show that smoothed analysis can overcome worst-case lower-bound barriers for distributed local graph problems.

0.09

Subgraph Counting with Local Node Differential Privacy

cs.DS cs.CR ·2026-10-15 ·Sofya Raskhodnikova, Adam Smith, Connor Wagaman, Anatoly Zavyalov
top venuetheory similarity 0.68
Abstract · To appear at the ACM-SIAM Symposium on Discrete Algorithms (SODA) 2027

We study the problem of approximately counting subgraphs under local node differential privacy (LNDP), a recently introduced model that enforces strong privacy guarantees in analyses of distributed social networks. We design a noninteractive $(\varepsilon,0)$-LNDP algorithm that, given any public $k$-node template graph $H$, counts occurrences of $H$ in a sensitive undirected $n$-node graph with additive error $O(n^{k-\frac{1}{2}}/\varepsilon^{k-1})$, for any constant $k$. We complement this algorithm with a lower bound showing that every noninteractive $(\varepsilon,δ)$-LNDP algorithm for counting copies of any $k$-node template graph with maximum degree greater than $1$ must have error $Ω(n^{k - \frac{1}{2}})$. Thus, our algorithm achieves optimal error for counting constant-size subgraphs, up to $\mathrm{poly}(1/\varepsilon)$ factors. Our results substantially expand the types of network statistics known to be approximable under LNDP: prior work considered only functions of the degree distribution. In addition, our lower bound yields the first known $\mathrm{poly}(n)$ gap between central and local node differential privacy when accuracy is required for all input graphs. Previous separations either only involved $\mathrm{poly}(1/\varepsilon)$ factors or only applied to promise problems.

0.32

Improved Quasi-Polynomial-Time Approximation for Maximum Independent Set

cs.DS cs.CC ·2026-10-26 ·Zhao Song
similarity 0.68
Abstract

We give a deterministic quasi-polynomial-time algorithm for the maximum independent set problem. If an $N$-vertex graph contains an independent set of size $2^{-t}N$, where $t$ is an integer with $1\le t\le\sqrt{\log N}/12$, then our algorithm finds an independent set of size $2^t$ in time $N^{O(t)}$. This gives an approximation ratio of $N/2^{Ω(\sqrt{\log N})}$ in time $N^{O(\sqrt{\log N})}$, improving on the approximation ratio $N/2^{Ω((\log N)^{1/3})}$ obtained by Lovett [Lov26]. A limitation of this form of analysis follows from the integrality gaps of Tulsiani [Tul09]. These gaps imply that a universal guarantee of the form $\operatorname{Las}_{R(t)}(G)\ge2^{-t}N\Longrightarrowα(G)\ge2^t$, at $2^{o(\sqrt{\log N\log\log N})}$ levels, cannot hold over a range extending beyond $t=O(\sqrt{\log N\log\log N})$. Thus, the exponent $1/2$ is the limit of this form of analysis, up to a factor of $\sqrt{\log\log N}$.

0.23

Tight Lower Bounds for Differentially Private Continual Counting

cs.DS cs.CR ·2026-09-15 ·Charlie Harrison, Ethan Leeman
theory similarity 0.66
Abstract

The Binary Tree Mechanism is a standard algorithm for differentially private continual counting, but its asymptotic optimality under pure differential privacy has remained unresolved since its introduction. We resolve this question. For fixed $0 < \varepsilon \le 1$, we prove asymptotically tight lower bounds of $Ω(\log^2 n)$ for worst-case expected $\ell_\infty$ error and $Ω(\log^3 n)$ for mean and maximum per-coordinate expected squared error. These bounds hold for arbitrary mechanisms, even when the entire stream is available in advance. The same lower bounds hold under approximate differential privacy whenever $δ\le n^{-c}$, for any fixed $c>0$. Our lower bounds match the Binary Tree Mechanism instantiated with Laplace noise, establishing its asymptotic optimality under both pure differential privacy and approximate differential privacy in the standard regime of $δ\ll1/n$. Our proof uses a single hard distribution with a bounded exponential score on a tree. A simple modification of the score allows the same framework to establish tight lower bounds for all three error measures.

0.42

Differentially Private All-Pairs Min Cut Values

cs.DS ·2026-11-05 ·Anders Aamand, Justin Y. Chen, Mina Dalirrooyfard, Slobodan Mitrović, Sandeep Silwal +1 more
publishedtheory similarity 0.66
Abstract · To appear in SOSA 2027

We study the problem of privately releasing the min cut values of all pairs of vertices in a given $n$-node weighted graph, under the edge-neighboring model. Straightforward application of advanced composition yields an approximate DP algorithm that outputs all min cut values up to $\tilde{O}(n)$ additive error. Prior works [Dalirrooyfard, Mitrovi{ć} and Nevmyvaka, 2023; Aamand et al., 2026] have established that $\tildeΘ(n)$ is the correct bound if the algorithm also needs to output cuts that realize the approximate min cut values. On the other hand, simply outputting the values could be an easier problem, but prior algorithms have been stuck on the $\tilde{O}(n)$ bound. In this work, we give an approximate DP algorithm outputting all-pair min cut values with $\tilde{O}(\sqrt{n})$ additive error, beating the linear error barrier. Our algorithm applies the sparse vector technique [Dwork and Roth, 2014] to maintain a hierarchical decomposition of the vertices based on the min cut structure. We also prove an $Ω(\log^{3/2} n)$ lower bound for this problem.

0.35