2026· International Conference on Security and Cryptography· pp. 625-636· 0 citations· 22 references
Computer Science
TL;DR
Across communication, citation, and social networks, the proposed method achieves competitive structural fidelity and downstream utility under user-level LDP, with particularly strong performance on assortativity and modularity at moderate to large privacy budgets.
Abstract
: Synthesising heterogeneous social graphs in decentralised settings is challenging because clients observe only egocentric views and strict privacy rules prevent central access to raw links or content. We present a locally private spectral graph synthesis framework for federated social platforms and peer-to-peer applications. Clients compute Laplacian spectral embeddings and degree summaries of their ego graphs and perturb them locally using the High-Dimensional Spherical mechanism for continuous embeddings and the two-sided Geometric mechanism for integer-valued degrees. An honest but curious server receives only the perturbed summaries and reconstructs a synthetic graph through Bayesian optimisation of reconstruction parameters. All server-side processing operates solely on noisy data and therefore incurs no additional privacy cost. Content is modelled explicitly as nodes, preserving homophily and heterophily. We evaluate structural fidelity through degree heterogeneity, clustering, assortativity, and modularity, and assess downstream utility through link prediction and community detection. Experiments are repeated under multiple random seeds and results are aggregated for robustness. Across communication, citation, and social networks, the proposed method achieves competitive structural fidelity and downstream utility under user-level LDP, with particularly strong performance on assortativity and modularity at moderate to large privacy budgets. Computational trade-offs, scalability considerations, and potential threats are also discussed.
We study hierarchical spectral graph clustering under edge differential privacy (DP) through the lens of iterative eigenvector estimation on adjacency matrices. We propose a differentially private recursive spectral framework, where each binary partition is obtained via a rank-one noisy power method applied to induced adjacency sub-matrices. At each iteration, carefully calibrated Gaussian noise is injected into the matrix–vector multiplication, ensuring (ε,δ)-edge DP under cumulative privacy accounting across both power iterations and recursive hierarchy levels while preserving the essential convergence properties of the classical power method. We provide a non-asymptotic analysis of the resulting noisy iterations, characterizing the trade-off between privacy and accuracy via explicit bounds on the eigenvector estimation error. In particular, we quantify how the noise variance, number of iterations, eigengap, and hierarchy depth jointly influence the accuracy of each recursive split and the overall clustering performance. Empirical evaluations on synthetic and real-world networks validate the theoretical predictions and demonstrate that the proposed method achieves strong multi-scale clustering performance under meaningful privacy budgets.
Mohamed Seif Eldin Mohamed, Andrea J. Goldsmith· Entropy· 0 citations
Graph neural networks learn from relational structure but can be sensitive to edge-level noise. We study a hybrid graph classifier that combines a message-passing branch (Graph Isomorphism Network, GIN) with a topological branch based on extended persistence diagrams and PersLay embeddings. Training optionally uses a paper-specific hinge penalty motivated by stable persistence-diagram representations and Lipschitz regularity, which we abbreviate as “HK-inspired.” On six TUDataset benchmarks, we compare five ablations (Full, GIN+PersLay, GIN + HK, GIN only, PersLay only) and measure robustness as accuracy drop under 10% random edge removal at test time, with persistence diagrams fixed from the original graphs; we also report targeted perturbations on MUTAG and PROTEINS and runtime on all six datasets. The regularized configurations reduce relative accuracy degradation where GIN-only is sensitive, particularly on smaller molecular and protein graphs, but do not consistently maximize clean or perturbed accuracy. On large social-network benchmarks, robustness differences are small and clean accuracy is the main differentiator. Adding PersLay to GIN can improve accuracy on several datasets, while PersLay alone performs worst. Training the full model increases per-epoch cost relative to GIN-only, with most overhead during training rather than inference. Overall, the results support a conditional empirical conclusion: combining structure, topology, and stability-oriented training can improve robustness under specified structural perturbations, with dataset-dependent accuracy–stability trade-offs rather than universal guarantees.
Jelena Losic, Charles Fanning· PLOS Complex Systems· 0 citations
Network embedding methods learn low-dimensional representations of graph-structured data to support downstream tasks such as node classification, link prediction, and influence maximization. However, real-world networks often reflect structural inequalities arising from demographic imbalances, homophily, and other societal biases, which fairness-agnostic embedding methods can encode and amplify. To address this issue, numerous fairness-aware network embedding methods have been proposed to mitigate bias while preserving embedding utility. This survey presents a comprehensive overview of fairness-aware network embeddings for complex networks. We propose a taxonomy that categorizes existing methods along three main complementary dimensions: underlying embedding approach (spectral, random walk, graph neural network, Bayesian, and method-agnostic), fairness intervention strategy (pre-processing, in-processing, and post-processing), and fairness objective criterion (embedding- or task-level). We further compare methods with respect to group versus individual fairness and assumptions regarding sensitive attributes. Finally, we discuss current limitations and highlight promising future research directions. This survey provides a unified perspective on fairness-aware network embedding and serves as a reference for developing fair and trustworthy network representation learning methods.
Ella Has, Harshit Yadav, Gaurav Dixit et al.· 0 citations
Experiments on four benchmarks for node classification and link prediction show that PriDyG consistently outperforms geometrically decaying baselines under the same privacy budget and matches the utility of naive per-update retraining while reducing cumulative privacy cost by up to three orders of magnitude.
Information diffusion prediction forecasts future participants from an observed cascade prefix, enabling proactive intervention in applications such as viral marketing and misinformation mitigation. Most existing models leverage two data sources: the global social graph (exposure/trust pathways) and cascade-induced interaction relations (interest-driven co-adoption), following a ''learn-then-fuse'' pipeline that encodes both graphs with GNNs and combines them via gated fusion to condition a sequential decoder. However, we find the two views are systematically mismatched: interaction edges are largely disjoint from social links, most social neighbors never co-activate within the same cascade, and the resulting embeddings lie on near-orthogonal manifolds with negligible correspondence. With such mismatch, static fusion is ill-posed: when the views disagree, fusion enforces a compromise and can cause negative interference. We further identify three reliability mechanisms that determine when each view should be trusted: (1) behavioral consensus across views is a high-fidelity signal of influence; (2) social cues are essential in cold-start regimes where interactions are sparse and biased; and (3) social ties dominate early seeding, while interaction patterns govern the late viral stage. Motivated by these, we propose CARD, a context-aware routing framework that replaces static fusion with step-wise evidence arbitration. CARD constructs an expert pool with social and interaction experts, a consensus expert that activates when both views are confirmed to behavioral consensus, and a graph-agnostic prior expert for noisy fallback. A router hard-selects the single most reliable expert at each step, so the decoder receives a targeted signal rather than a blurred mixture. Extensive experiments on four real-world datasets show that CARD achieves state-of-the-art accuracy and stronger robustness.
Zihan Feng, Yajun Yang, Rui Wu et al.· Proceedings of the 32nd ACM...· 0 citations
Orthogonal Decomposition for Social Recommendation (ODSR) is proposed, an embedding-space framework that orthogonally decomposes the aggregated social message into an aligned component and an orthogonal deviation, and learns a dimension-wise vector gate to regulate the deviation under ranking supervision.
Rongfeng Guo, Yinxuan Huang, Wei Chen et al.· Proceedings of the 32nd ACM...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
