Reproduce “Beyond Attention: Learning Spatio-Temporal Dynamics with Emergent Interpretable Topologies” 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
Sai Vamsi Alisetti, Vikas Kalagi, Sanjukta Krishnagopal
InterGAT-GRU forecasts traffic speeds by learning a symmetric node-to-node interaction matrix shared across time, using it to aggregate information before a GRU models temporal changes. On two traffic datasets, it reports lower MAE at all four horizons than GAT-GRU; at 60 minutes on SZ-Taxi, MAE is 2.9652 versus 4.6701. These results are limited to two relatively small datasets.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.