search

pyt-team / challenge-icml-2024

Official repository for the Topological Deep Learning Challenge 2024, organized by TAG-DS & PyT-Team and hosted by GRaM Workshop @ ICML 2024.
https://pyt-team.github.io/packs/challenge.html
MIT License
37 stars 48 forks source link

Node Attribute Lifting #39

Closed demiqin closed 2 months ago

demiqin commented 2 months ago

This method involves lifting a graph to a hypergraph based on node attributes. Hyperedges are created by grouping nodes with the same attribute. Given (n) attributes of a node, users can choose which attribute to use for constructing the hyperedges. Additionally, users can preprocess the data to add new attributes for grouping. For example, in a social network, hyperedges can represent friend circles, or nodes can be grouped based on domain-specific distances.

This lifting approach is particularly useful for datasets where node attributes are present. For instance, the MUTAG, ENZYMES, and PROTEINS datasets all have this property. We tested our method on these datasets and updated the tutorial accordingly.

review-notebook-app[bot] commented 2 months ago

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB