HyperNetwork Approximating Future Parameters for Time Series Forecasting Under Temporal Drifts
Abstract
Models for time series forecasting require the ability to extrapolate from previous observations. Yet, extrapolation is challenging, especially when the data spanning several periods is under temporal drifts where each period has a different distribution. To address this problem, we propose HyperGPA, a hypernetwork that generates a target model's parameters that are expected to work well (i.e., be an optimal model) for each period. HyperGPA discovers an underlying hidden dynamics which causes temporal drifts over time, and generates the model parameters for a target period, aided by the structures of computational graphs. In comprehensive evaluations, we show that target models whose parameters are generated by HyperGPA are up to 64.1\% more accurate than baselines.
Cite
Text
Lee et al. "HyperNetwork Approximating Future Parameters for Time Series Forecasting Under Temporal Drifts." NeurIPS 2023 Workshops: DistShift, 2023.Markdown
[Lee et al. "HyperNetwork Approximating Future Parameters for Time Series Forecasting Under Temporal Drifts." NeurIPS 2023 Workshops: DistShift, 2023.](https://mlanthology.org/neuripsw/2023/lee2023neuripsw-hypernetwork/)BibTeX
@inproceedings{lee2023neuripsw-hypernetwork,
title = {{HyperNetwork Approximating Future Parameters for Time Series Forecasting Under Temporal Drifts}},
author = {Lee, Jaehoon and Kim, Chan and Lee, Gyumin and Lim, Haksoo and Choi, Jeongwhan and Lee, Kookjin and Lee, Dongeun and Hong, Sanghyun and Park, Noseong},
booktitle = {NeurIPS 2023 Workshops: DistShift},
year = {2023},
url = {https://mlanthology.org/neuripsw/2023/lee2023neuripsw-hypernetwork/}
}