LeapTS: Rethinking Time Series Forecasting as Adaptive Multi-Horizon Scheduling
2026-05-11 • Machine Learning
Machine LearningArtificial Intelligence
AI summaryⓘ
The authors introduce LeapTS, a new way to make time series forecasts by treating prediction as a step-by-step scheduling process rather than a fixed mapping from past data. LeapTS uses a hierarchical controller to choose how far and at what scale to predict next, combined with neural controlled differential equations to model changes over continuous time. This approach helps the model adjust predictions dynamically as new information comes in, leading to better accuracy and faster inference compared to traditional Transformer models. The authors also show that LeapTS can adapt its forecasting behavior to handle changes in the data over time.
time series forecastinghierarchical controllerneural controlled differential equationsdynamic schedulingnon-stationary dynamicsTransformer modelscontinuous-time modelinginference speedup
Authors
Sheng Pan, Ming Jin, Bo Du, Shirui Pan
Abstract
Time series forecasting serves as an essential tool for many real-world applications, supporting tasks such as resource optimization and decision-making. Despite significant architectural advancements, most modern models still treat forecasting task as a fixed mapping from history to target horizons. This induces temporal decoupling across future time points and limits the model's ability to adapt to the evolving context as forecasting progresses. In this work, we present LeapTS, a novel framework that reformulates time series forecasting as a dynamic scheduling process over the prediction horizon. Specifically, LeapTS organizes the forecasting process into multi-level decisions using: (1) the hierarchical controller to dynamically select the optimal prediction scale and advancement length at each step, and (2) continuous-time state evolution driven by neural controlled differential equations. Within this process, the controlled update mechanism explicitly couples the irregular temporal dynamics with discrete scheduling feedback. Extensive evaluations on both real-world and synthetic datasets demonstrate that LeapTS improves overall forecasting performance by at least 7.4% while achieving a 2.6$\times$ to 5.3$\times$ inference speedup over representative Transformer-based models. Furthermore, by explicitly tracing the scheduling trajectories, we reveal how the model autonomously adapts its forecasting behavior to capture non-stationary dynamics.