FiLSim scales graph similarity with GNNs
Original title: Scaling Graph Similarity Learning with Graph Neural Networks
Framework scores 150,000-node graphs in seconds while baselines fail.
Graph similarity computation using Graph Edit Distance and Maximum Common Subgraph addressed with GNNs. Existing methods limited to small datasets. FiLSim framework proposed with sequential graph-editing strategy for large-scale datasets.
- Handles graphs over 150,000 nodes
- Scores test pairs in seconds on single GPU
- Four benchmark and four large-scale datasets
- GNN pipeline maintains competitive performance
- Scalable beyond baseline memory limits
From the original article
1 min read at arxiv.orgGraph similarity computation is an important task in graph mining with many real-world applications, such as chemical compound analysis, program dependence comparison, and social network analysis.
By Ahmet Tüzen · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.