Gradient-Based Spectral Embeddings of Random Dot Product Graphs

被引:0
作者
Fiori, Marcelo [1 ,2 ]
Marenco, Bernardo [1 ,2 ]
Larroca, Federico [1 ]
Bermolen, Paola [1 ,2 ]
Mateos, Gonzalo [3 ]
机构
[1] Univ Republica, Facultadde Ingn, Montevideo 11000, Uruguay
[2] Univ Republica, Ctr Interdisciplinario Ciencia Datos & Aprendizaje, Montevideo 11000, Uruguay
[3] Univ Rochester, Dept Elect & Comp Engn, Rochester, NY 14627 USA
来源
IEEE TRANSACTIONS ON SIGNAL AND INFORMATION PROCESSING OVER NETWORKS | 2024年 / 10卷
关键词
Optimization; Information processing; Matrix decomposition; Manifolds; Representation learning; Data models; Scalability; Graph representation learning; gradient descent; manifold optimization; random dot product graphs; MATRIX FACTORIZATION;
D O I
10.1109/TSIPN.2023.3343607
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
The Random Dot Product Graph (RDPG) is a generative model for relational data, where nodes are represented via latent vectors in low-dimensional Euclidean space. RDPGs crucially postulate that edge formation probabilities are given by the dot product of the corresponding latent positions. Accordingly, the embedding task of estimating these vectors from an observed graph is typically posed as a low-rank matrix factorization problem. The workhorse Adjacency Spectral Embedding (ASE) enjoys solid statistical properties, but it is formally solving a surrogate problem and can be computationally intensive. In this paper, we bring to bear recent advances in non-convex optimization and demonstrate their impact to RDPG inference. We advocate first-order gradient descent methods to better solve the embedding problem, and to organically accommodate broader network embedding applications of practical relevance. Notably, we argue that RDPG embeddings of directed graphs loose interpretability unless the factor matrices are constrained to have orthogonal columns. We thus develop a novel feasible optimization method in the resulting manifold. The effectiveness of the graph representation learning framework is demonstrated on reproducible experiments with both synthetic and real network data. Our open-source algorithm implementations are scalable, and unlike the ASE they are robust to missing edge data and can track slowly-varying latent positions from streaming graphs.
引用
收藏
页码:1 / 16
页数:16
相关论文
共 44 条
[1]  
Absil PA, 2008, OPTIMIZATION ALGORITHMS ON MATRIX MANIFOLDS, P1
[2]  
Athreya A, 2018, J MACH LEARN RES, V18
[3]  
Belkin P., 2001, ADV NEURAL INF PROCE, P1
[4]  
Bertsekas D.P., 1999, Nonlinear programming, V2nd
[5]  
Bhojanapalli S., 2016, C LEARN THEOR, P530
[6]  
Boumal N ..., 2023, An Introduction to Optimization on Smooth Manifolds
[7]   Global rates of convergence for nonconvex optimization on manifolds [J].
Boumal, Nicolas ;
Absil, P-A. ;
Cartis, Coralia .
IMA JOURNAL OF NUMERICAL ANALYSIS, 2019, 39 (01) :1-33
[8]   Fast low-rank modifications of the thin singular value decomposition [J].
Brand, M .
LINEAR ALGEBRA AND ITS APPLICATIONS, 2006, 415 (01) :20-30
[9]   Time-aware recommender systems: a comprehensive survey and analysis of existing evaluation protocols [J].
Campos, Pedro G. ;
Diez, Fernando ;
Cantador, Ivan .
USER MODELING AND USER-ADAPTED INTERACTION, 2014, 24 (1-2) :67-119
[10]  
Capdehourat G, 2020, 2020 IFIP NETWORKING CONFERENCE AND WORKSHOPS (NETWORKING), P370