Reproduce “To Learn is to Wander: Learning Across Graphs and Tasks with Random Walks” 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
Louis Tichelman, Xingyue Huang, Jinwoo Kim, İsmail İlkan Ceylan
Wander uses a pretrained model to predict missing node labels and edges across homogeneous and knowledge graphs. It represents graph context with sampled random walks, adjustable at inference without changing model parameters. Across benchmarks it reports aggregate results. On the Grids test, accuracy rises from 56.6% to 100% with longer, targeted walks; this does not establish performance on other graphs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.