VMFTransformer: An Angle-Preserving and Auto-Scaling Machine for Multi-horizon Probabilistic Forecasting

Yunyi Zhou, Zhixuan Chu, Ruohan Gao, Xinping Zheng, Yuchen Huang, Sheng Li

Primary Area: general machine learning (i.e., none of the above)

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Keywords: time series forecasting, probabilistic forecasting

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

Abstract: Time series forecasting has historically been a key area of academic research and industrial applications. As deep learning develops, the major research methodologies of time series forecasting can be divided into two categories, i.e., iterative and direct methods. In the iterative methods, since a small amount of error is produced at each time step, the recursive structure can potentially lead to large error accumulations over longer forecasting horizons. Although the direct methods can avoid this puzzle involved in the iterative methods, it faces abuse of conditional independence among time points. This impractical assumption can also lead to biased models. To solve these challenges, we propose a direct approach for multi-horizon probabilistic forecasting, which can effectively characterize the dependence across future horizons. Specifically, we consider the multi-horizon target as a random vector. The direction of the vector embodies the temporal dependence, and the length of the vector measures the overall scale across each horizon. Therefore, we respectively apply the von Mises-Fisher (VMF) distribution and the truncated normal distribution to characterize the angle and the magnitude of the target vector in our model. We evaluate the performance of our framework on three benchmarks. Extensive results demonstrate the superiority of our framework over six state-of-the-art methods and show the remarkable versatility and extensibility for different time series forecasting tasks.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 8832