Private Edge Density Estimation for Random Graphs: Optimal, Efficient and Robust

We give the first polynomial-time, differentially node-private, and robust algorithm for estimating the edge density of Erdős-Rényi random graphs and their generalization, inhomogeneous random graphs. We further prove information-theoretical lower bounds, showing that the error rate of our algorithm is optimal up to logarithmic factors. Previous algorithms incur either exponential running time or suboptimal error rates. Two key ingredients of our algorithm are (1) a new sum-of-squares algorithm for robust edge density estimation, and (2) the reduction from privacy to robustness based on sum-of-squares exponential mechanisms due to Hopkins et al. (STOC 2023).

Paper

Similar papers

Peer review

Reviewer 97Rf6/10 · confidence 3/52024-06-28

Summary

This paper studies edge density estimation for Erdos-Renyi random graph under node-level differential privacy. They then show that their approach can actually be used to estimating edge density for inhomogeneous graphs. In particular, this paper proposes an efficient algorithm with optimal privacy cost $O(1/(\varepsilon n \sqrt{np}))$, which is negligible to the non-private error for any epsilon in a moderate regime. To achieve this, the main technical approach in this paper is to design a new sum-of-squares algorithm for robust edge density estimation, and then based on the reduction from privacy to robustness in Hopkins et al., this paper uses an exponential mechanism whose score function is based on the correspoinding the sum-of-squares program. To design a robust algorithm for edge density estimation of ER random graphs, this paper established several polynomial constraints to ensure that a ER random graph will satisfy with high probability. The proof that if a graph meets these constraints, its average degree will remain close to $d$ is straightforward enough for the sum-of-squares proof system, extends the utility guarantee of the polynomial program to its semidefinite programming relaxation, resulting in a polynomial-time robust algorithm.

Strengths

1. Graph parameter approximation under node-level privacy is an important topic, and this paper gives optimal algorithm with a polynomial time implementation by sum-of-squares relaxation, under the most natrual setting of random graphs. 2. Their techniques of using sum-of-squares method for exponential mechanism is sophisticated. 3. All theorems are clearly stated and technically correct.

Weaknesses

As far as I can see, the most relevant work is Chen et al. "Private graphon estimation via sum-of-squares", but it appears to me that this paper lacks discussion and comparison in [Chen et al.]. Please see "Questions" for details. I would be happy to raise my score if all my questions are adressed.

Questions

1. Could you please elaborate on which part of your proof technique deviates from that in [Chen et al.]? Alternatively, are the proof techniques similar, while you are just addressing different problems in this paper? 2. There is also a theorem in [Chen et al.] for edge density estimation of stochastic block random graphs (Lemma 4.10), and it appears that their utility is better. Could you explain the fundamental obstacle between stochastic block random graphs and the more general inhomogeneous graphs that prevents the estimation for inhomogeneous graphs from achieving results as good as those for stochastic block random graphs? Would you mind adding more discussion about the results in [Chen et al.]?

Rating

6

Confidence

3

Soundness

3

Presentation

3

Contribution

2

Limitations

This paper discusses several limitations.

Reviewer Se7o7/10 · confidence 3/52024-07-12

Summary

The authors propose a robust and efficient (polynomial time) algorithm to estimate the edge density of Erdos-Renyi graph, which achieves optimal error up to logarithmic factors. The authors also design an optimal algorithm for inhomogenous graphs where edges are drawn iid from different Bern(p_i).

Strengths

1. The theoretical results are very strong. The authors prove matching (up to logarithmic factors) error bounds for both Inhomogeneous and Erdos-Renyi graphs. The algorithms are polynomial-time. 2. The techniques are interesting. The authors use recent results for the connection between privacy and robustness. To design a time-efficient algorithm, the authors go beyond simple reduction from privacy to robustness and apply the sum-of-squares method. 3. The paper is well-written.

Weaknesses

There is no obvious weakness with regard to the results and technical parts of the paper. The only minor weaknesses are the problem formulation. 1. Both the Erdos-Renyi graph and the inhomogeneous model assume edges are independently chosen. In real applications, graphs are often fixed, and edges may not be independent. 2. The algorithm only works for edge density estimation. There should be many other interesting graph statistics, even on the Erdos-Renyi graph.

Questions

1. Is it possible to design algorithms when the entries of the connection probability matrix are not independent? 2. Do any algorithmic ideas translate to other graph statistics like the number of triangles and k-stars?

Rating

7

Confidence

3

Soundness

3

Presentation

3

Contribution

3

Limitations

Limitations and negative social impact are adequately addressed.

Reviewer teKp7/10 · confidence 2/52024-07-13

Summary

This paper gives the first polynomial time DP algorithm for estimating edge density of random graphs (Erdos-Renyi and inhomogeneous). The authors also give information-theoretic lower bounds to show that the error achieved is optimal (up to log factors). The paper utilizes the recent results of Hopkins et al who gave a black box reduction from privacy to robustness via a sum-of-squares exponential mechanism to design their new algorithm. Their main new contribution is a sum-of-squares algorithm for estimating the edge density which they then use together with the Hopkins et al framework to get their final algorithm.

Strengths

The main result of the paper is novel and significantly advances the area of DP algorithms for estimating graph parameters in random graphs.

Weaknesses

The techniques used heavily rely on the recent framework given by the Hopkins et al STOC paper. What is unclear to me is --- what were the challenges in designing the sum-of-squares algorithm for estimating the edge density? Since this algorithm is the main contribution of the paper in some sense, I would like to understand better if the design of the sum-of-squares algorithm itself was previously known and one only needed the Hopkins et al framework to obtain this result. General comment about the paper --- you should define the quantity edge density formally somewhere since the whole paper hinges on the reader understanding what it is.

Questions

The results of this paper are very important and exciting. However, can you expand on the challenges in designing the sum-of-squares algorithm for estimating the edge density? I would like to be more convinced that this result is not simply piggybacking off on the Hopkins et al result. Thanks!

Rating

7

Confidence

2

Soundness

3

Presentation

3

Contribution

3

Limitations

There was no explicit discussion regarding the limitations of this work in the main body.

Reviewer ZgW78/10 · confidence 3/52024-07-15

Summary

This paper presents a sum-of-squares-based polynomial-time differentially node-private algorithm for estimating the edge density of Erdos-Renyi random graphs that achieves optimal accuracy; the algorithm is simultaneously robust to corruptions. Specifically, if p is the true Erdos-Renyi parameter, then the estimate q released by the algorithm has error |q/p - 1| bounded by 1/(n sqrt{p}) + 1/(eps n \sqrt{np}) + eta / sqrt{np} (up to factors of log n), where eta is the corruption rate for robustness. Up to log n factors, the first term is the information-theoretically optimal error rate with neither privacy or robustness (which is also the error of the empirical edge density), and the third term is known to be required for robustness ([AJK+22]). The paper further shows a lower bound demonstrating that the remaining error term is necessary for privacy (a lower bound had been shown by BCSZ18 for a closely related family of random graphs but not for standard Erdos-Renyi graphs, so the new lower bound closes this gap). Moreover, the paper extend this result to present an algorithm for the much more general problem of edge density estimation in inhomogeneous random graphs (e.g. SBMs or graphons) and proves lower bounds in this setting as well showing that this algorithm also achieves optimal error rate up to logarithmic factors.

Strengths

This paper achieves optimal accuracy in polynomial time for both private and robust edge density estimation of Erdos-Renyi graphs and for the more general setting of inhomogeneous random graphs, nicely closing the gaps from prior work. It achieves this by bringing a new approach to the problem: the sum-of-squares framework, along with leveraging a reduction from privacy to robustness. The paper is generally well-written. For DP edge density estimation of Erdos-Renyi graphs (even without requiring robustness), the new algorithm matches the (already optimal) accuracy achieved by [BCSZ18] while running in polynomial time instead of exponential time; it also improves over the rate of the polynomial-time algorithm of [SU19], which was suboptimal in the sparse or very-high-privacy regimes (that is, when epsilon * sqrt{pn} << 1). For robust edge density estimation of Erdos-Renyi graphs (even without requiring privacy), the new algorithm also improves of the accuracy of prior work ([AJK+22]) in the sparse regime. Thus, the Erdos-Renyi result improves on SOTA for both privacy and robustness, achieving the optimal bound (up to log n factors). Moreover, the results are extended to the more challenging inhomogenous random graph setting, achieving optimal rate in polynomial time here as well, and novel, tight lower bounds are shown for both settings as well.

Weaknesses

The discussion of prior work is clear for the Erdos-Renyi setting, but it would be useful to have a clearer discussion of what was previously known for the more general inhomogeneous random graph setting under DP and under robustness; see the specific questions 1--2 below.

Questions

1) How does the rate achieved for inhomogeneous random graphs compare to the rate achieved by the exponential-time algorithm of [BCSZ18] for graphons? 2) Is there prior work on robust estimation of general inhomogeneous random graphs, and if so, what rate was achieved? 3) Can we hope to extend the privacy lower bounds (Theorem 1.5 and 1.8) to (epsilon, delta)-DP and not just epsilon-DP? Minor comment: line 188: should d^\circ be defined as (n-1) p^\circ instead of n p^\circ?

Rating

8

Confidence

3

Soundness

4

Presentation

4

Contribution

4

Limitations

Yes, adequately addressed.

Reviewer ikor6/10 · confidence 3/52024-07-17

Summary

The paper concerns the graph density estimation question for random Erdos-Renyi graphs. The optimal algorithm, without additional constraints, for this problem is to output the density of edges in the graph. The paper considers a node-private robust flavor of the question, however: 1. The output is not supposed to reveal much information about any node. More precisely, for any node, the distribution on estimates the algorithm produces is not supposed to be significantly different whether it is included or not. (This task is much easier to achieve in the edge privacy model.) 2. The algorithm is supposed to be robust, i.e., if the edges involving a small number of vertices are arbitrarily edited, this should not throw the estimate off significantly. (This may require, for instance, disregarding some fraction of vertices that are significantly off.) The paper gives a near optimal algorithm in this setting that runs in polynomial time and it establishes asymptotic bounds on the error achieved by the algorithm. Additionally, the paper gives an extension of the result to non-homogeneous random graphs.

Strengths

This is a technically impressive achievement with the paper combining many desirable properties of algorithms for a natural toy problem. This is an interesting result for anyone studying private estimation and private release of graph parameters.

Weaknesses

This is not directly a very practical problem, since Erdos-Renyi graphs do not appear frequently in natural contexts and even the generalization assumes that the selection of different edges is independent, which is not true for many models of real-world graphs. The algorithm may be difficult to implement and run in practice due to the complexity of tools used, which include semidefinite programming.

Questions

Just to make sure, in the inhomogeneous setting, the algorithm has to know $Q^\circ$, right? Otherwise, it could be a zero-one matrix equal to the graph adjacency matrix. What guarantees can be given if $Q^\circ$ is unknown?

Rating

6

Confidence

3

Soundness

4

Presentation

3

Contribution

3

Limitations

No limitations.

Reviewer 97Rf2024-08-11

Thank you for the rebuttal.

Program Chairsdecision2024-09-25

Decision

Accept (spotlight)

© 2026 NYSGPT2525 LLC