Network Embedding as Matrix Factorization: Unifying DeepWalk, LINE, PTE, and node2vec
Explore this paper's citation graph
Summary
The NetMF method offers significant improvements over DeepWalk and LINE for conventional network mining tasks and provides the theoretical connections between skip-gram based network embedding algorithms and the theory of graph Laplacian.
- Type
- preprint
- Published
- 2017-10-09
- Cited by
- 994
- References
- 58
- Access
- Open access
- OpenAlex
- https://openalex.org/W2761896323
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:3952914
Keywords
Theoretical computer science, Matrix decomposition, Computer science, Embedding, Factorization
References
- PathSim
- Link Mining: Models, Algorithms, and Applications
- Efficient Sampling for Gaussian Graphical Models via Spectral Sparsification
- ARPACK users' guide - solution of large-scale eigenvalue problems with implicitly restarted Arnoldi methods
- Networks, Crowds and Markets: Reasoning about a Highly Connected World
- Tensor Spectral Clustering for Partitioning Higher-order Network Structures
- Numerical linear algebra
- Spectral Graph Theory
- Efficient Estimation of Word Representations in Vector Space
- Comprehend DeepWalk as Matrix Factorization
- LINE: Large-scale Information Network Embedding
- Learning latent representations of nodes for classifying in heterogeneous social networks
- Feature-Rich Part-of-Speech Tagging with a Cyclic Dependency Network
- Relational learning via latent social dimensions
- Heterogeneous Network Embedding via Deep Architectures
- GraRep: Learning Graph Representations with Global Structural Information
- Empirical comparison of algorithms for network community detection
- Using the Nyström Method to Speed Up Kernel Machines
- LIBLINEAR: A Library for Large Linear Classification
- Neural Word Embedding as Implicit Matrix Factorization
Cited by
- Biological Systems as Heterogeneous Information Networks: A Mini-review and Perspectives
- Semi-Supervised Learning on Graphs Based on Local Label Distributions
- GEMSEC: Graph Embedding with Self Clustering
- MILE: A Multi-Level Framework for Scalable Graph Embedding
- Node Representation Learning for Multiple Networks: The Case of Graph Alignment
- Models for Capturing Temporal Smoothness in Evolving Networks for Learning Latent Representation of Nodes
- CARL: Content-Aware Representation Learning for Heterogeneous Networks
- Spectral Network Embedding: A Fast and Scalable Method via Sparsity
- A topic-based cross-language retrieval model with PLSA and TF-IDF
- MASTER: across Multiple social networks, integrate Attribute and STructure Embedding for Reconciliation
- Discrete Network Embedding
- Adversarially Regularized Graph Autoencoder
- Arbitrary-Order Proximity Preserved Network Embedding
- Content to Node: Self-Translation Network Embedding
- High-order Proximity Preserving Information Network Hashing
- DeepInf: Social Influence Prediction with Deep Learning
- Multi-task Representation Learning for Travel Time Estimation
- Measuring Graph Reconstruction Precisions: How Well Do Embeddings Preserve the Graph Proximity Structure?
- Efficient Training on Very Large Corpora via Gramian Estimation
- User Tagging in MOOCs Through Network Embedding
Related papers
- GraRep: Learning Graph Representations with Global Structural Information
- Inductive Representation Learning on Large Graphs
- Semi-Supervised Classification with Graph Convolutional Networks
- metapath2vec: Scalable Representation Learning for Heterogeneous Networks
- Structural Deep Network Embedding
- Asymmetric Transitivity Preserving Graph Embedding