Node embedding dimensionality impacts stability, new arXiv paper finds

A new arXiv paper explores how the dimensionality of node embeddings impacts their stability across different training runs and methods. Researchers found that dimensionality can significantly affect stability, with effects varying by embedding method and stability metric. For instance, node2vec and ASNE generally became more stable at higher dimensions, while GraphSAGE and VERSE showed more complex or decreasing stability. The study also noted that dimensions yielding high stability do not always correlate with the best downstream performance. AI

IMPACT Provides insights into optimizing node embedding stability for improved downstream performance in graph-based machine learning tasks.

RANK_REASON Academic paper published on arXiv detailing research findings. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• ASNE
• GraphSAGE
• Node2vec
• Tobias Schumacher

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →