Pith. sign in

REVIEW 2 major objections 4 minor 3 cited by

Change-point detection in dynamic networks via graphon estimation

T0 review · 2 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A two-stage graphon-smoothing procedure consistently recovers the number and locations of change-points in dynamic networks.

desk verdict A solid methodological extension of neighborhood smoothing to dynamic networks with a real, fixable gap in the proof of the main consistency theorem. read the letter →

arxiv 1908.01823 v1 pith:WBTHEQCV submitted 2019-08-05 stat.ME math.STstat.TH

Zifeng Zhao , Li Chen , Lizhen Lin This is my paper · ORCID
classification stat.MEmath.STstat.TH MSC 62G2062H1262M10
keywords change-pointdetectiondynamicnetworksgraphonestimationneighborhoodsmoothinglinkprobabilitymatrixscanstatisticsconsistencystochasticblockmodel
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proposes a two-stage procedure for detecting when the link structure of a network changes over time. First, it estimates the underlying link probability matrix in each time window using a modified neighborhood smoothing (MNBS) estimator that averages adjacency matrices across time while shrinking the smoothing neighborhood. Second, it scans local windows, computes a $d_{2,\infty}$ distance between the two estimated matrices, and thresholds local maximizers. The paper proves that, under a piecewise Lipschitz graphon model, the procedure consistently recovers the true number and window-localized positions of all change-points, and that the estimation error decreases with both the number of nodes and the number of temporal snapshots. The practical payoff is a model-free detector that exploits network structure rather than collapsing networks to scalar summaries.

What carries the argument

The central object is the modified neighborhood smoothing (MNBS) estimator, a temporal extension of neighborhood smoothing: for each node $i$, form a neighborhood $N_i$ of nodes with similar row-wise connection patterns in the averaged adjacency matrix $\bar{A} = (1/T)\sum_t A^{(t)}$, using a distance measure $\tilde{d}(i,i') = \max_{k \neq i,i'} |\langle \bar{A}_{i\cdot} - \bar{A}_{i'\cdot}, \bar{A}_{k\cdot} \rangle|$, then estimate $P_{ij}$ by averaging $\bar{A}_{i'j}$ over $i' \in N_i$. The key modification is shrinking the neighborhood quantile $q$ from $C(\log n/n)^{1/2}$ to $C\log n/(n^{1/2}\omega)$, trading increased variance (compensated by averaging over $T$ snapshots) for reduced bias. The detection stage uses the $d_{2,\infty}$ distance between MNBS estimates from windows before and after each time $t$, followed by thresholding of local maximizers.

What would settle it

Generate a dynamic network from a stochastic block model where node memberships gradually drift across time (e.g., a small fraction of nodes switch blocks between segments), or where adjacency matrices are generated with first-order temporal dependence at fixed $P$; then run the MNBS procedure with $h=\sqrt{T}$ and check empirically whether the estimated change-point set still satisfies the sure-coverage event. A clean calculation would also verify the threshold condition: if $\Delta_*/\Delta_D \le 1$, the proof's key inequality fails and the procedure should lose detection power, so a simulation at the boundary $\Delta_*/\Delta_D$ just below 1 provides a direct test.

Watch

Extended reading notes

Core claim

The paper establishes that change-point detection in a dynamic network can be reduced to estimating link probability matrices and then scanning for differences. The MNBS estimator pools $T$ adjacency matrices, shrinking the neighborhood size from $q \sim (\log n/n)^{1/2}$ in the single-network NBS to $q \sim \log n/(n^{1/2}\omega)$ with $\omega = \min(n^{1/2}, (T\log n)^{1/2})$, achieving $d_{2,\infty}(\tilde{P}, P)^2 = O(\log n/(n^{1/2}\omega))$. The detection procedure then computes $D(t,h) = d_{2,\infty}(\tilde{P}_{t1,h}, \tilde{P}_{t2,h})^2$ on a sliding window of length $h$, collects $h$-local maximizers, and keeps those above a threshold $\Delta_D$ of order $(\log n)^{1/2+\delta_0}/(n^{1/2}h^{1/2})$. Theorem 4.1 states that if $h < D_*/2$ (the minimal segment length halved) and the minimum signal $\Delta_*$ exceeds the threshold asymptotically, then $P(\{\hat{J} = J\} \cap \{J \subset \hat{J} \pm h\}) \to 1$; if the signal is strong enough, setting $h=1$ recovers exact change-point locations, in contrast to the classical $O_p(1)$ localization error for scalar time series.

Load-bearing premise

The proof assumes that within each time segment every edge probability is a piecewise Lipschitz function of fixed, i.i.d. latent node positions, and that network snapshots are independent across time; if node roles drift over time or snapshots are temporally dependent, the bias and variance bounds, and with them the detection consistency proof, collapse.

Editorial extensions

If this is right

  • When nodes outnumber time points ($n > T$), the detection rate improves with both $n$ and $T$, beating methods that average the network into a single time series.
  • Exact localization (zero error) becomes possible when the change signal is strong enough, unlike classical time-series change-point detection where localization error is $O_p(1)$.
  • The procedure is model-free within the graphon class, covering Erdős–Rényi and stochastic block models as special cases.
  • The same MNBS estimates provide post-hoc network estimation, so detection and estimation come from one pipeline.
  • The separation condition is stated in $d_{2,\infty}$ norm, which is weaker than Frobenius-norm separation used in prior work, so changes affecting few nodes (e.g., one node switching membership) remain detectable.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the graphon model is right, the principle of shrinking smoothing neighborhoods in proportion to available temporal samples could extend to other dynamic-network tasks such as community detection or link prediction, where repeated observations should reduce bias rather than only variance.
  • The fixed, i.i.d. latent-position assumption is a natural stress point; a testable extension would allow small, smooth drift in latent positions between segments and ask how much drift the $d_{2,\infty}$ separation condition can tolerate before consistency breaks.
  • The rates are derived for undirected networks; adapting MNBS to directed or weighted networks would require redefining the distance measure and the bias-variance decomposition, but the screening-thresholding structure would likely survive.
  • The window constraint $h < D_*/2$ and the practical recommendation $h = \sqrt{T}$ mean that segments shorter than $2\sqrt{T}$ may be missed; a sensitivity analysis of the $h$-versus-$D_*$ trade-off would clarify the procedure's practical limits.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 4 minor

Summary. The paper proposes a multiple change-point detection procedure for dynamic networks, built on a modified neighborhood smoothing (MNBS) estimator for the link probability matrix. Section 3 introduces MNBS for repeated observations of a network and proves an error bound (Theorem 3.2) under a piecewise Lipschitz graphon model. Section 4 develops a screening-and-thresholding algorithm based on local MNBS estimates over windows of size h, with a detection threshold ΔD, and states a consistency result (Theorem 4.1): under h < D*/2 (D* the minimum segment length) and lim Δ*/ΔD > 1, the procedure satisfies P({Ĵ = J} ∩ {J ⊂ Ĵ ± h}) → 1, i.e., exact recovery of the number of change-points and window-localized estimation of their positions. Numerical experiments on synthetic stochastic block model dynamics and an MIT proximity network illustrate the method's practical performance, with comparisons to graph-based scan methods.

Significance. If the main results are correct, the paper makes a valuable contribution to change-point detection in dynamic networks: it proposes a nonparametric approach that exploits network structure, provides a graphon-estimation error rate that improves with both n and T, and uses a d_{2,∞}-based signal measure that is weaker than Frobenius-norm conditions used in earlier work. The theorems are stated in a parameter-free way (holding for any positive tuning constants B0, D0, δ0), which is a strength, and the numerical comparisons show competitive or superior performance relative to existing methods. However, the central consistency theorem (Theorem 4.1) has a proof gap concerning the uniqueness of local maximizers, which is load-bearing for the claimed exact recovery of the change-point set; the stated signal condition is insufficient as written for the conclusion to follow.

major comments (2)
  1. [Section 9, proof of Theorem 4.1] The event ξ_n only controls D(t,h) at exact change-points τ ∈ J and at h-flat points t ∈ F_h. It does not control D(τ±k,h) for 1 ≤ k < h, because those indices are not h-flat. Consequently, the proof does not rule out multiple h-local maximizers above ΔD within the same h-neighborhood of a true change-point. The argument establishes that each estimated point has exactly one true change-point within distance h and each true change-point has at least one estimated point within distance h, but it does not establish that each true change-point has at most one estimated point. The claimed implication ξ_n ⇒ {Ĵ = J} therefore does not follow. The issue is not merely technical: under the stated condition lim Δ*/ΔD > 1, the deterministic signal near a true change-point decays as (1 − k/h)^2 Δ*, so D(τ,h) − D(τ+1,h) is of order Δ*/h, while the estimation noise in D is of order √(Δ* err) with err = (log n)^{1/2}/(n^{1/2} h^{1/2}). The noise-to-signal ratio for this decrement is of order h (log n)^{−δ0/2}, which diverges when h grows faster than (log n)^{δ0/2}. Thus, for large h, the event ξ_n does not preclude additional local maxima above ΔD near a single true change-point, and the conclusion {Ĵ = J} is not established by the given proof. A stronger signal condition (e.g., Δ*/ΔD growing with h) or a new argument controlling D(t,h) on the entire h-neighborhood is needed.
  2. [Section 4.1 and Theorem 4.1] The theorem and the algorithm description lack explicit boundary assumptions. The scan statistic D(t,h) is defined only for t = h, ..., T−h, but the definition of an h-local maximizer in Section 4.1 requires comparing D(x,h) with D(t,h) for all t in [x−h+1, x+h−1]; for x near the boundaries, this interval includes indices where D(t,h) is not defined. Similarly, Theorem 4.1 does not state that the true change-points lie in [h, T−h], which is necessary for the windows used in the proof to be fully observed. These conditions should be made explicit, and the proof should address the boundary cases or exclude them by assumption.
minor comments (4)
  1. [Section 9, proof of Theorem 4.1] In the concluding paragraph of the proof, “for any point ˆτ∈J” should read “for any point ˆτ∈Ĵ”, since J denotes the true change-point set and Ĵ the estimated set.
  2. [Theorem 4.1] The notation “J⊂: Ĵ±h” is confusing; suggest writing the coverage condition explicitly as: for each j = 1, ..., J, there exists ␣τ_j ∈ Ĵ with |␣τ_j − τ_j| < h, and |Ĵ| = J.
  3. [Section 5.1] The text says “The performance of MNBS and CZ are summarized in Table 6”, but the displayed table in the main text is Table 1; the cross-reference should be corrected.
  4. [Abstract and Section 3] The abstract claims a faster convergence rate in detecting change-points compared with an algorithm that simply averages information across time, but Theorem 4.1 provides only window-consistency with no rate for localization error, and no formal detection-rate comparison with an averaged-information detector is given. The claim is supported for graphon estimation (Theorem 3.2) but not for change-point detection; consider rewording or supplying a formal rate statement.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the consistency theorems are parameter-free derivations from explicit graphon assumptions, and the only self-citation is not load-bearing.

full rationale

The paper's derivation chain is self-contained rather than circular. Theorem 3.2 (Consistency of MNBS) is proved directly using Bernstein inequalities and the piecewise Lipschitz graphon assumption via Lemmas 9.2 and 9.3, which are explicitly stated extensions of Lemmas 1 and 2 in the external prior work [39] (Zhang, Levina, and Zhu). The target rate is not assumed; it is obtained from bias-variance bounds. Theorem 4.1 imposes explicit conditions (h < D*/2 and lim Delta*/Delta_D > 1) and then proves that, under the event xi_n, flat points are excluded and each true change-point has an h-local maximizer above threshold. The threshold Delta_D is chosen to dominate the estimation error of Theorem 3.2, and the theorem is stated for any positive constants B0, D0, delta0, so no fitted parameter is renamed as a prediction. The only self-citation is [38] (Yau and Zhao), used as one example of a screening-and-thresholding strategy for classical time series; it is not used to justify the network-specific consistency claim or to forbid alternative methods. The skeptical concern about the possible non-uniqueness of h-local maximizers in the proof of Theorem 4.1 is a potential correctness gap, not a circularity: it questions whether the stated assumptions imply the event, not whether the theorem reduces to its inputs by construction. The numerical benchmarks compare MNBS against the external graph-based test of [7] (Chen and Zhang), so the empirical claim is not validated by re-expression of the method's own outputs. Overall, no definitional equivalence, fitted-input-as-prediction, or author-imported uniqueness argument appears in the central derivation.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the graphon model, piecewise Lipschitz smoothness, independence of snapshots, and separation and signal conditions. The only hand-chosen inputs are the four tuning parameters, all of which are proven to work for any positive values, so they do not compromise the theoretical claim.

free parameters (4)
  • h = sqrt(T) (recommended)
    Local window size for the scan statistic; any h < D*/2 works in theory, recommended h = √T to balance localization and signal detection.
  • B0 = 3 (recommended, with 1 or 2 similar)
    Multiplier in the neighborhood quantile q = B0 (log n)^{1/2}/(n^{1/2} h^{1/2}); the theorem holds for any B0 > 0.
  • D0 = 0.25 (recommended)
    Multiplier in the detection threshold ΔD; the theorem holds for any D0 > 0.
  • delta0 = 0.1 (recommended)
    Exponent in ΔD = D0 (log n)^{1/2+δ0}/(n^{1/2} h^{1/2}); the theorem holds for any δ0 > 0 and needs δ0 > 0 for the threshold to dominate the estimation error.
assumptions (5)
  • standard math Bernstein inequality and standard concentration bounds
    Used throughout the supplementary proofs.
  • domain assumption Latent variables ξ_i are i.i.d. Uniform[0,1] and shared across time
    Section 3: P_ij = f(ξ_i, ξ_j); this is the exchangeability assumption for the graphon model.
  • domain assumption The link probability matrix in each segment is generated by a piecewise Lipschitz graphon from F_{δ;L} with common δ and L
    Definition 3.1 and Section 4.2; needed for the neighborhood approximation Lemma 9.3.
  • domain assumption Adjacency matrices are independent across time and conditionally on P(t) are independent Bernoulli entries
    Sections 3 and 4; the variance reduction from T relies on this.
  • domain assumption Change-points are separated by at least 2h (h < D*/2) and the minimum signal exceeds the threshold (Δ*/ΔD > 1)
    Assumptions in Theorem 4.1; needed for the union bound over flat points and for detection of true change-points.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Change-point detection in dynamic networks via graphon estimation." pith.science (2026). https://pith.science/paper/WBTHEQCV

@misc{pith2026190801823,
  author       = {Pith},
  title        = {Pith review of: Change-point detection in dynamic networks via graphon estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WBTHEQCV}},
  note         = {Machine review of arXiv:1908.01823}
}
read the original abstract

We propose a general approach for change-point detection in dynamic networks. The proposed method is model-free and covers a wide range of dynamic networks. The key idea behind our approach is to effectively utilize the network structure in designing change-point detection algorithms. This is done via an initial step of graphon estimation, where we propose a modified neighborhood smoothing~(MNBS) algorithm for estimating the link probability matrices of a dynamic network. Based on the initial graphon estimation, we then develop a screening and thresholding algorithm for multiple change-point detection in dynamic networks. The convergence rate and consistency for the change-point detection procedure are derived as well as those for MNBS. When the number of nodes is large~(e.g., exceeds the number of temporal points), our approach yields a faster convergence rate in detecting change-points comparing with an algorithm that simply employs averaged information of the dynamic network across time. Numerical experiments demonstrate robust performance of the proposed algorithm for change-point detection under various types of dynamic networks, and superior performance over existing methods is observed. A real data example is provided to illustrate the effectiveness and practical impact of the procedure.

Figures

Figures reproduced from arXiv: 1908.01823 by the authors.

Figure 1
Figure 1. Average number of estimated change-points [PITH_FULL_IMAGE:figures/full_fig_p013_1.png] view at source ↗
Figure 2
Figure 2. Scan statistics, local-maximizers and estimated change-points by MNBS for MIT network [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Dynamic Networks with Node Heterogeneity and Homophily

    math.ST 2026-08 conditional novelty 7.0 of 10

    A dynamic network model jointly estimating node heterogeneity and observed plus latent homophily, with a normalized squared loss and consistency theory for high-dimensional node-specific parameters.

  2. Hierarchical Clustering of Networks via Hierarchical Distance Matrices

    stat.ME 2026-07 conditional novelty 7.0 of 10

    A provably consistent top-down procedure that recovers latent hierarchical clusters of networks from hierarchical distance matrices.

  3. A Proof of The Changepoint Detection Threshold Conjecture in Preferential Attachment Models

    math.PR 2025-02 accept novelty 7.0 of 10

    Changepoint detection in preferential attachment networks is impossible when the change occurs in the last o(√n) steps, resolving the Bet-Castro-van der Hofstad conjecture.

Reference graph

Works this paper leans on

47 extracted references · 43 canonical work pages · cited by 3 Pith papers

  1. [1]

    On semidefinite relaxations for the block model

    Arash A Amini and Elizaveta Levina. On semidefinite relaxations for the block model. arXiv preprint arXiv:1406.5647, 2014

    arXiv 2014

  2. [2]

    Efficient and principled method for detecting communities in networks

    Brian Ball, Brian Karrer, and MEJ Newman. Efficient and principled method for detecting communities in networks. Physical Review E, 84(3):036103, 2011

    work page 2011

  3. [3]

    Change point estimation in a dynamic stochastic block model

    Monika Bhattacharjee, Moulinath Banerjee, and George Michailidis. Change point estimation in a dynamic stochastic block model. arXiv:1812.03090v1, 2018

    arXiv 2018

  4. [4]

    Spectral Clustering for Multiple Sparse Networks: I

    Sharmodeep Bhattacharyya and Shirshendu Chatterjee. Spectral Clustering for Multiple Sparse Networks: I. arXiv e-prints, page arXiv:1805.10594, May 2018

    arXiv 2018

  5. [5]

    Consistencies and rates of convergence of jump-penalized least squares estimators

    Leif Boysen, Angela Kempe, V olkmar Liebscher, Axel Munk, and Olaf Wittich. Consistencies and rates of convergence of jump-penalized least squares estimators. The Annals of Statistics, 37(1):157–183, 2009. 9

    work page 2009

  6. [6]

    Matrix estimation by universal singular value thresholding

    Sourav Chatterjee. Matrix estimation by universal singular value thresholding. Ann. Statist., 43(1):177–214, 02 2015

    work page 2015

  7. [7]

    Graph-based change-point detection

    Hao Chen and Nancy Zhang. Graph-based change-point detection. Ann. Statist., 43(1):139–176, 02 2015

    work page 2015

  8. [8]

    Multiple change points detection and clustering in dynamic networks

    Marco Corneli, Pierre Latouche, and Fabrice Rossi. Multiple change points detection and clustering in dynamic networks. Statistics and Computing, 28(5):989–1007, 2017

    work page 2017

Show all 47 references
  1. [9]

    Eagle, A

    N. Eagle, A. Pentland, and D. Lazer. Inferring social network structure using mobile phone data. Proceedings of the National Academy of Sciences, 106(36):15274–15278, 2009

    2009

  2. [10]

    Chao Gao, Yu Lu, and Harrison H. Zhou. Rate-optimal graphon estimation. Ann. Statist., 43(6):2624–2652, 12 2015

    2015

  3. [11]

    C. E. Ginestet, P. Balanchandran, S. Rosenberg, and E. D. Kolaczyk. Hypothesis Testing For Network Data in Functional Neuroimaging. ArXiv e-prints, July 2014

    2014

  4. [12]

    Stochastic blockmodels: First steps

    Paul W Holland, Kathryn Blackmond Laskey, and Samuel Leinhardt. Stochastic blockmodels: First steps. Social networks, 5(2):109–137, 1983

    1983

  5. [13]

    Brian Karrer and M. E. J. Newman. Stochastic blockmodels and community structure in networks. Phys. Rev. E, 83:016107, Jan 2011

    2011

  6. [14]

    Averages of Unlabeled Networks: Geometric Characterization and Asymptotic Behavior

    Eric Kolaczyk, Lizhen Lin, Steven Rosenberg, Jie Xu, and Jackson Walters. Averages of Unlabeled Networks: Geometric Characterization and Asymptotic Behavior. arXiv e-prints, page arXiv:1709.02793, Sep 2017

    2017 arXiv

  7. [15]

    Deltacon: Principled massive-graph similarity function with attribution

    Danai Koutra, Neil Shah, Joshua T V ogelstein, , Brian Gallagher, and Christos Faloutsos. Deltacon: Principled massive-graph similarity function with attribution. ACM Transactions on Knowledge Discovery from Data (TKDD), 10(3), 2016

    2016

  8. [16]

    C.B. Lee. Nonparametric multiple change-point estimators. Statistics and Probability Letters, 27:295–304, 1996

    1996

  9. [17]

    A tutorial on spectral clustering.Statistics and Computing, 17(4):395–416, 2007

    Ulrike V on Luxburg. A tutorial on spectral clustering.Statistics and Computing, 17(4):395–416, 2007

    2007

  10. [18]

    Sequential changepoint approach for online commu- nity detection

    David Marangoni-Simonsen and Yao Xie. Sequential changepoint approach for online commu- nity detection. IEEE Signal Processing Letters, 22(8):1035–1039, 2015

    2015

  11. [19]

    Ian McCulloh and Kathleen M. Carley. Detecting change in longitudinal social networks. Journal of Social Structure, 12, 2011

    2011

  12. [20]

    Network hypothesis testing using mixed kronecker product graph models

    Sebastian Moreno and Jennifer Neville. Network hypothesis testing using mixed kronecker product graph models. 2013 IEEE 13th International Conference on Dataing Mining, pages 1163–1168, 2013

    2013

  13. [21]

    S. S. Mukherjee. On some inference problems for networks.Doctoral dissertation, UC Berkeley, 2018

    2018

  14. [22]

    On clustering network-valued data

    Soumendu Sundar Mukherjee, Purnamrita Sarkar, and Lizhen Lin. On clustering network-valued data. In I. Guyon, U. V . Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, editors, Advances in Neural Information Processing Systems 30, pages 7071–7081. Curr...

    2017

  15. [23]

    A. Ng, M. Jordan, and Y . Weiss. On spectral clustering: Analysis and an algorithm. InAdvances in Neural Information Processing Systems, 2001

    2001

  16. [24]

    Niu and Heping Zhang

    Yue S. Niu and Heping Zhang. The screening and ranking algorithm to detect dna copy number variations. Annals of Applied Statistics, 6(3):1306–1326, 2012

    2012

  17. [25]

    Detecting change points in the large-scale structure of evolving networks

    Leto Peel and Aaron Clauset. Detecting change points in the large-scale structure of evolving networks. AAAI’15 Proceedings of the Twenty-Ninth AAAI Conference on Artificial Intelligence, pages 2914–2920, 2015. 10

    2015

  18. [26]

    Dynamic network models and graphon estimation

    Marianna Pensky. Dynamic network models and graphon estimation. arXiv:1607.00673v2, 2018

    2018 arXiv

  19. [27]

    Priebe, John M

    Carey E. Priebe, John M. Conroy, David J. Marchette, and Youngser Park. Scan statistics on enron graphs. Computational & Mathematical Organization Theory, 11(3):229–247, 2005

    2005

  20. [28]

    A. E. Raftery, M. S. Handcock, and P. D. Hoff. Latent space approaches to social network analysis. Journal of the American Statistical Association, 15:460, 2002

    2002

  21. [29]

    Samatova

    Stephen Ranshous, Shitian Shen, Danai Koutra, Steve Harenberg, Christos Faloutsos, and Nagiza F. Samatova. Anomaly detection in dynamic networks: A survey.WIREs Computational Statistics, 7(3):223–247, 2015

    2015

  22. [30]

    Spectral clustering and the high-dimensional stochastic block model

    Karl Rohe, Sourav Chatterjee, and Bin Yu. Spectral clustering and the high-dimensional stochastic block model. Annals of Statistics, 39:1878–1915, 2011

    1915

  23. [31]

    Sarkar and A

    P. Sarkar and A. Moore. Dynamic social network analysis using latent space models. In Advances in Neural Information Processing Systems. 2005

    2005

  24. [32]

    Sewell and Yuguo Chen

    Daniel K. Sewell and Yuguo Chen. Latent space models for dynamic networks. Journal of the American Statistical Association, 110(512):1646–1657, 2015

    2015

  25. [33]

    disorder

    L. Ju. V ostrikova. Detecting “disorder” in multidimensional random processes.Soviet Mathe- matics: Doklady, 24:55–59, 1981

    1981

  26. [34]

    Optimal change point detection and localization in sparse dynamic networks

    Daren Wang, Yi Yu, and Alessandro Rinaldo. Optimal change point detection and localization in sparse dynamic networks. arXiv:1809.09602v1, 2018

    2018 arXiv

  27. [35]

    Fast change point detection on dynamic social networks

    Yu Wang, Aniket Chakrabarti, David Sivakoff, and Srinivasan Parthasarathy. Fast change point detection on dynamic social networks. Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence (IJCAI-17), pages 2992–2998, 2017

    2017

  28. [36]

    White and P

    S. White and P. Smyth. A spectral clustering approach to finding communities in graphs. In SIAM Intl. Conf. on Data Mining, 2005

    2005

  29. [37]

    Approximating the distribution of the maximum likelihood estimate of the change-point in a sequence of independent random variables

    Yi-Ching Yao. Approximating the distribution of the maximum likelihood estimate of the change-point in a sequence of independent random variables. Annals of Statistics, 15(3):1321– 1328, 1987

    1987

  30. [38]

    Yau and Z

    C.Y . Yau and Z. Zhao. Inference for multiple change-points in time series via likelihood ratio scan statistics. Journal of the Royal Statistical Society - Series B, 78(4):895–916, 2016

    2016

  31. [39]

    Estimating network edge probabilities by neighbour- hood smoothing

    Yuan Zhang, Elizaveta Levina, and Ji Zhu. Estimating network edge probabilities by neighbour- hood smoothing. Biometrika, 104(4):771–783, 2017

    2017

  32. [40]

    Nonparametric maximum likelihood approach to multiple change-point problems

    Changliang Zou, Guosheng Yin, Long Feng, and Zhaojun Wang. Nonparametric maximum likelihood approach to multiple change-point problems. The Annals of Statistics, 42(3):970– 1002, 2014. 7 Supplementary material: additional numerical studies Section 7.1 defines additional SBMs/gr...

    2014

  33. [41]

    Let ¯A =∑n i=1siuiuT i be the singular value decomposition of the average adjacent matrix ¯A

  34. [42]

    Let ˆ¯A =∑ i∈SsiuiuT i

    LetS ={i : si≥ (2 +η) √n√ T}, where η∈ (0, 1) is some small positive number. Let ˆ¯A =∑ i∈SsiuiuT i

  35. [43]

    ˆP serves as the final estimate for P

    Let ˆP = (ˆPij), where ˆPij :=    ˆ¯Aij, if 0≤ˆ¯Aij≤ 1 1, ifˆ¯Aij≥ 1 0, ifˆ¯Aij≤ 0. ˆP serves as the final estimate for P . The key distinction between our estimate and the one in [ 6] is that we utilize ¯A, which allows us to lower the threshold level from an order of √n ...

  36. [44]

    Proof of Lemma 9.3

    With probability 1− 2n−( ˜C1+γ)− 2n−( ˜C2+γ), for alli andi′∈N i, we have ‖Pi·−Pi′·‖2 2/n≤ (6LC1 + 24C3) logn n1/2ω. Proof of Lemma 9.3. The first claim follows immediately from the definition of quantile and q, since |Ni|≥ n·q =nB0 logn n1/2ω =B0 n1/2 logn ω . To prove the seco...

  37. [45]

    Lemma 9.2: n1/2 ω · (C1−B1)2 7C1−B1 >γ + 1 andC1 >B 1 > 0; 22

  38. [46]

    Lemma 9.3: C2 3 logn 6 · T ω2 > 2 +γ and C3 2 · n1/2 ω > 2 +γ andB1≥B0

  39. [47]

    C2 4 (logn)3 4 · n ω2· T 2 ω2 > (1 +γ) and 3C4 logn 4 · n ω2 > 1 +γ; C2 5 logn 6 · T 2 ω2 > 2 +γ and C5 2 · n1/2 ω > 2 +γ; ω2 T (logn)2≤ 1. It is easy to see that, for any γ >0 andB0 > 0, we can always find B1,C 1,C 3,C 4,C 5 such that all inequalities in (1)-(3) hold for alln ...

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.