Reproduce “Geometric Matrix Completion with Recurrent Multi-Graph Neural Networks” 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
Federico Monti, Michael M. Bronstein, Xavier Bresson
RGCNN learns to complete missing user-item ratings by extracting patterns from user and item graphs, then repeatedly updating predictions with a recurrent network. Its learned graph-filter coefficients do not grow with matrix dimensions, though computation still scales with matrix size. Full RGCNN reports 0.0053 RMSE on synthetic data; on MovieLens, sRGCNN reports 0.929 versus GRALS at 0.945, among listed methods.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.