GOPHER: Categorical probabilistic forecasting with graph structure via local continuous-time dynamics

Ke A. Wang, Danielle Maddix, Yuyang Wang
Proceedings on "I (Still) Can't Believe It's Not Better!" at NeurIPS 2021 Workshops, PMLR 163:80-85, 2022.

Abstract

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.

Cite this Paper

BibTeX
@InProceedings{pmlr-v163-wang22a, title = {{GOPHER}: Categorical probabilistic forecasting with graph structure via local continuous-time dynamics}, author = {Wang, Ke A. and Maddix, Danielle and Wang, Yuyang}, booktitle = {Proceedings on "I (Still) Can't Believe It's Not Better!" at NeurIPS 2021 Workshops}, pages = {80--85}, year = {2022}, editor = {Pradier, Melanie F. and Schein, Aaron and Hyland, Stephanie and Ruiz, Francisco J. R. and Forde, Jessica Z.}, volume = {163}, series = {Proceedings of Machine Learning Research}, month = {13 Dec}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v163/wang22a/wang22a.pdf}, url = {https://proceedings.mlr.press/v163/wang22a.html}, abstract = {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.} }
Endnote
%0 Conference Paper %T GOPHER: Categorical probabilistic forecasting with graph structure via local continuous-time dynamics %A Ke A. Wang %A Danielle Maddix %A Yuyang Wang %B Proceedings on "I (Still) Can't Believe It's Not Better!" at NeurIPS 2021 Workshops %C Proceedings of Machine Learning Research %D 2022 %E Melanie F. Pradier %E Aaron Schein %E Stephanie Hyland %E Francisco J. R. Ruiz %E Jessica Z. Forde %F pmlr-v163-wang22a %I PMLR %P 80--85 %U https://proceedings.mlr.press/v163/wang22a.html %V 163 %X 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.
APA
Wang, K.A., Maddix, D. & Wang, Y.. (2022). GOPHER: Categorical probabilistic forecasting with graph structure via local continuous-time dynamics. Proceedings on "I (Still) Can't Believe It's Not Better!" at NeurIPS 2021 Workshops, in Proceedings of Machine Learning Research 163:80-85 Available from https://proceedings.mlr.press/v163/wang22a.html.

Related Material