Reproduce “Set Twister for Single-hop Node Classification” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yangze Zhou, Vinayak Rao, Bruno Ribeiro
Set Twister predicts node labels from a node’s features and its immediate neighbors’ features, combining pooled feature channels to represent interactions among neighbors. Unlike DeepSets, it enriches the intermediate representation while retaining linear scaling in neighborhood size for fixed settings. On PubMed, it reaches 90.15% test accuracy, but results vary across datasets, so one-hop is not established as generally sufficient.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.