Research map / Graph learning / Graph algorithms and clustering
DeepWalking Backwards: From Embeddings Back to Graphs
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use gradient-based optimization instead of the analytical inversion approach | NetMF embedding inversion · Real-world network benchmark | relative Frobenius error between reconstructed and original low-rank PPMI matric | Helped |
| Increase embedding dimension up to a certain point | Optimization-based NetMF inversion · Real-world networks | relative Frobenius error between reconstructed and original adjacency matrices | Helped |
| Increase embedding dimension | Optimization-based NetMF inversion · Real-world networks | triangle-count relative error | Mixed |
| Reconstruct a graph by optimizing its PPMI to match the input embedding | NetMF · Real-world networks | community conductance | Helped |
| Reconstruct a graph by optimizing its PPMI to match the input embedding | NetMF · Real-world networks | average path length relative error | Helped |
| Use embeddings of the reconstructed real-valued graph rather than the original graph | NetMF node classification · Real-world networks | micro F1 | No clear effect |
| Binarize the reconstructed graph before generating embeddings | NetMF node classification · Real-world networks | micro F1 | Hurt |
| Use the full-rank PPMI matrix rather than a low-dimensional embedding | NetMF inversion · Stochastic block model networks | adjacency reconstruction | Helped |
| Use the analytical inversion approach under finite window size and reduced embedding dimension | NetMF embedding inversion · Real-world network benchmark | relative Frobenius error between reconstructed and original low-rank PPMI matric | Hurt |
Closest papers by meaning
- On the Optimization Trajectory of DeepWalk Embeddings (ICML 2026 · similarity 0.90)
- How Low Can You Go? Searching for the Intrinsic Dimensionality of Complex Networks using Metric Node Embeddings (ICLR 2025 · similarity 0.89)
- Learning Representations for Hierarchies with Minimal Support (NeurIPS 2024 · similarity 0.89)
- Information Loss and Disparate Effects in Network Embeddings (NeurIPS 2026 · similarity 0.88)
- Weighted Embeddings for Low-Dimensional Graph Representation (AAAI 2025 · similarity 0.88)
- Convergence Guarantees for the DeepWalk Embedding on Block Models (ICML 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.