Taming the Long Tail of Deep Probabilistic Forecasting
This addresses a critical limitation in forecasting applications like weather, electricity consumption, and autonomous vehicles where rare events matter, representing an incremental improvement through novel loss functions.
The paper tackles the problem of poor performance on rare and difficult cases in deep probabilistic forecasting by identifying a long tail behavior in state-of-the-art methods and proposing two moment-based tailedness measurement concepts (Pareto Loss and Kurtosis Loss). The approach achieves significant improvements on tail examples across several real-world datasets including time series and spatiotemporal trajectories.
Deep probabilistic forecasting is gaining attention in numerous applications ranging from weather prognosis, through electricity consumption estimation, to autonomous vehicle trajectory prediction. However, existing approaches focus on improvements on the most common scenarios without addressing the performance on rare and difficult cases. In this work, we identify a long tail behavior in the performance of state-of-the-art deep learning methods on probabilistic forecasting. We present two moment-based tailedness measurement concepts to improve performance on the difficult tail examples: Pareto Loss and Kurtosis Loss. Kurtosis loss is a symmetric measurement as the fourth moment about the mean of the loss distribution. Pareto loss is asymmetric measuring right tailedness, modeling the loss using a generalized Pareto distribution (GPD). We demonstrate the performance of our approach on several real-world datasets including time series and spatiotemporal trajectories, achieving significant improvements on the tail examples.