Predictive Whittle Networks for Time Series

Zhongjie Yu, Fabrizio Ventola, Nils Thoma, Devendra Singh Dhami, Martin Mundt, Kristian Kersting

Keywords: Time Series, Forecasting, Probabilistic Circuits, Spectral Modeling, Whittle Networks

TL;DR: We propose Predictive Whittle Networks that perform more accurate and trustworthy time series forecasting by weighting the predictions of a spectral neural forecaster with the likelihoods obtained from a tractable probabilistic model.

Abstract: Recent developments have shown that modeling in the spectral domain improves the accuracy in time series forecasting. However, state-of-the-art neural spectral forecasters do not generally yield trustworthy predictions. In particular, they lack the means to gauge predictive likelihoods and provide uncertainty estimates. We propose predictive Whittle networks to bridge this gap, which exploit both the advances of neural forecasting in the spectral domain and leverage tractable likelihoods of probabilistic circuits. For this purpose, we propose a novel Whittle forecasting loss that makes use of these predictive likelihoods to guide the training of the neural forecasting component. We demonstrate how predictive Whittle networks improve real-world forecasting accuracy, while also allowing a transformation back into the time domain, in order to provide the necessary feedback of when the model's prediction may become erratic.