# Efficient Synthetic Network Generation via Latent Embedding Reconstruction Source: [Efficient Synthetic Network Generation via Latent Embedding Reconstruction]() Collection index: [llms.txt]() Data updated: 2026-06-02T05:00:51.558Z Authors: Feifan Jiang, Yinan Bu, Shihao Wu, Gongjun Xu, Ji Zhu Published: 2026-05-31T00:01:13.000Z ## Abstract Network data are ubiquitous across the social sciences, biology, and information systems. Generating realistic synthetic network data has broad applications from network simulation to scientific discovery. However, many existing black-box approaches for network generation tend to overfit observed data while overlooking characteristic network structure, and incur substantial computational overhead at scale. These practical challenges call for synthetic network generation methods that are both efficient and capable of capturing structural properties of networks. In this paper, we introduce Synthetic Network Generation via Latent Embedding Reconstruction (SyNGLER), a general and efficient framework for synthetic network generation that builds on latent space network models. Given an observed network, SyNGLER first learns low-dimensional latent node embeddings via a latent space network model and then reconstructs the latent space by building a distribution-free generator over these embeddings. For generation, SyNGLER first samples (or resamples) node embeddings from the generator in the latent space and then produces synthetic networks using the latent space network model. Through the latent space framework, SyNGLER preserves unique characteristics in networks such as sparsity and node degree heterogeneity, while allowing for efficient training with lower computational cost than many existing deep architectures. We provide theoretical guarantees by developing consistency results on the distance between the true and synthetic edge distributions. Empirical studies further demonstrate the effectiveness of SyNGLER, which efficiently produces networks that better preserve key network characteristics such as network moments and degree distributions compared with existing approaches. Code is available at https://github.com/FeifanJiang/syngler. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Synthetic Network Generation on PolBlogs](): Synthetic Network Generation; PolBlogs; GFDL1 Score: 0.72. - [Synthetic Network Generation on YouTube](): Synthetic Network Generation; YouTube; Eigenvalue MMD: 0.38. - [Synthetic Network Generation on PolBlogs (Triangle, Clustering, Eigenvalue, Degree Centrality)](): Synthetic Network Generation; PolBlogs; Triangle Similarity (MMD): 0.59. - [Synthetic Attributed Network Generation on Cora](): Synthetic Attributed Network Generation; Cora; KS Distance: 0.1225. - [Synthetic Network Generation on DBLP (Density and Clustering Error)](): Synthetic Network Generation; DBLP; Triangle Density Error: 1.43. - [Synthetic Network Generation on Simulated sparse networks n=500, r=2](): Synthetic Network Generation; Simulated sparse networks n=500, r=2; Triangle RMSE: 4.6. - [Synthetic Network Generation on Simulated sparse networks n=1500, r=4](): Synthetic Network Generation; Simulated sparse networks n=1500, r=4; Triangle RMSE: 0.63. - [Synthetic Network Generation on YouTube (Triangle Density and Clustering Coefficient Error)](): Synthetic Network Generation; YouTube; Triangle Density Error: 0.97. - [Synthetic Network Generation on Sparse Simulated Networks n=500, r=2](): Synthetic Network Generation; Sparse Simulated Networks n=500, r=2; Triangle RMSE: 4.6. - [Synthetic Network Generation on PolBlogs (Density and Clustering Error)](): Synthetic Network Generation; PolBlogs; Triangle Density Error: 0.71. - [Node Classification on Cora (GCN and MLP Results)](): Node Classification; Cora; GCN Performance: 0.98. - [Synthetic Network Generation on DBLP (Link Prediction Utility)](): Synthetic Network Generation; DBLP; Link Prediction Accuracy: 100. - [Synthetic Network Generation on Sparse Simulated Networks n=500, r=3](): Synthetic Network Generation; Sparse Simulated Networks n=500, r=3; Triangle RMSE: 7.31. - [Synthetic Network Generation on Sparse Simulated Networks n=500, r=4](): Synthetic Network Generation; Sparse Simulated Networks n=500, r=4; Triangle RMSE: 17.22. - [Synthetic Network Generation on PolBlogs (Orbit and 4-Cyc)](): Synthetic Network Generation; PolBlogs; Orbit: 0.0002. - [Synthetic Network Generation on DBLP (Orbit and 4-Cyc Metrics)](): Synthetic Network Generation; DBLP; Orbit: 0.035. - [Synthetic Network Generation on YouTube (Orbit and 4-Cyc)](): Synthetic Network Generation; YouTube; Orbit: 0.2. - [Synthetic Network Generation on YouTube (Link Prediction Accuracy)](): Synthetic Network Generation; YouTube; Link Prediction Accuracy Ratio: 99. - [Synthetic Network Generation on DBLP](): Synthetic Network Generation; DBLP; Transitivity: 3.63. - [Synthetic Network Generation on Yelp friendship network](): Synthetic Network Generation; Yelp friendship network; Triangle Ratio (10^-4): 2. - [Synthetic Network Generation on DBLP (GFDL Metrics)](): Synthetic Network Generation; DBLP; GFDL1: 9.51. - [Synthetic Network Generation on YouTube (GFDL1, GFDL2)](): Synthetic Network Generation; YouTube; GFDL1 Score: 3.02. - [Synthetic Network Generation on Yelp](): Synthetic Network Generation; Yelp; GFDL1 Link Prediction Accuracy: 0.42. ## Original sources and implementation [Original paper]() [PDF]()