Neural Stochastic Differential Games for Time-series Analysis

Sungwoo Park, Byoungwoo Park, Moontae Lee, Changhee Lee
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:27269-27293, 2023.

Abstract

Modeling spatiotemporal dynamics with neural differential equations has become a major line of research that opens new ways to handle various real-world scenarios (e.g., missing observations, irregular times, etc.). Despite such progress, most existing methods still face challenges in providing a general framework for analyzing time series. To tackle this, we adopt stochastic differential games to suggest a new philosophy of utilizing interacting collective intelligence in time series analysis. For the implementation, we develop the novel gradient descent-based algorithm called deep neural fictitious play to approximate the Nash equilibrium. We theoretically analyze the convergence result of the proposed algorithm and discuss the advantage of cooperative games in handling noninformative observation. Throughout the experiments on various datasets, we demonstrate the superiority of our framework over all the tested benchmarks in modeling time-series prediction by capitalizing on the advantages of applying cooperative games. An ablation study shows that neural agents of the proposed framework learn intrinsic temporal relevance to make accurate time-series predictions.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-park23j, title = {Neural Stochastic Differential Games for Time-series Analysis}, author = {Park, Sungwoo and Park, Byoungwoo and Lee, Moontae and Lee, Changhee}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {27269--27293}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/park23j/park23j.pdf}, url = {https://proceedings.mlr.press/v202/park23j.html}, abstract = {Modeling spatiotemporal dynamics with neural differential equations has become a major line of research that opens new ways to handle various real-world scenarios (e.g., missing observations, irregular times, etc.). Despite such progress, most existing methods still face challenges in providing a general framework for analyzing time series. To tackle this, we adopt stochastic differential games to suggest a new philosophy of utilizing interacting collective intelligence in time series analysis. For the implementation, we develop the novel gradient descent-based algorithm called deep neural fictitious play to approximate the Nash equilibrium. We theoretically analyze the convergence result of the proposed algorithm and discuss the advantage of cooperative games in handling noninformative observation. Throughout the experiments on various datasets, we demonstrate the superiority of our framework over all the tested benchmarks in modeling time-series prediction by capitalizing on the advantages of applying cooperative games. An ablation study shows that neural agents of the proposed framework learn intrinsic temporal relevance to make accurate time-series predictions.} }
Endnote
%0 Conference Paper %T Neural Stochastic Differential Games for Time-series Analysis %A Sungwoo Park %A Byoungwoo Park %A Moontae Lee %A Changhee Lee %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-park23j %I PMLR %P 27269--27293 %U https://proceedings.mlr.press/v202/park23j.html %V 202 %X Modeling spatiotemporal dynamics with neural differential equations has become a major line of research that opens new ways to handle various real-world scenarios (e.g., missing observations, irregular times, etc.). Despite such progress, most existing methods still face challenges in providing a general framework for analyzing time series. To tackle this, we adopt stochastic differential games to suggest a new philosophy of utilizing interacting collective intelligence in time series analysis. For the implementation, we develop the novel gradient descent-based algorithm called deep neural fictitious play to approximate the Nash equilibrium. We theoretically analyze the convergence result of the proposed algorithm and discuss the advantage of cooperative games in handling noninformative observation. Throughout the experiments on various datasets, we demonstrate the superiority of our framework over all the tested benchmarks in modeling time-series prediction by capitalizing on the advantages of applying cooperative games. An ablation study shows that neural agents of the proposed framework learn intrinsic temporal relevance to make accurate time-series predictions.
APA
Park, S., Park, B., Lee, M. & Lee, C.. (2023). Neural Stochastic Differential Games for Time-series Analysis. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:27269-27293 Available from https://proceedings.mlr.press/v202/park23j.html.

Related Material