GOPHER: Categorical probabilistic forecasting with graph structure via local continuous-time dynamics
This addresses probabilistic forecasting for structured data like social networks or biological systems, but it is incremental as it builds on existing graph and ODE methods.
The paper tackled probabilistic forecasting over categories with graph structure by proposing GOPHER, which combines graph neural networks and neural ODEs to capture local continuous-time dynamics, finding that graph structure improves accuracy and sample efficiency, but continuous-time evolution provided little benefit.
We consider the problem of probabilistic forecasting over categories with graph structure, where the dynamics at a vertex depends on its local connectivity structure. We present GOPHER, a method that combines the inductive bias of graph neural networks with neural ODEs to capture the intrinsic local continuous-time dynamics of our probabilistic forecasts. We study the benefits of these two inductive biases by comparing against baseline models that help disentangle the benefits of each. We find that capturing the graph structure is crucial for accurate in-domain probabilistic predictions and more sample efficient models. Surprisingly, our experiments demonstrate that the continuous time evolution inductive bias brings little to no benefit despite reflecting the true probability dynamics.