GatedLinear: Adaptive Routing of Complementary Linear Bases for Time Series Forecasting
2026-07-10 • Machine Learning
Machine Learning
AI summaryⓘ
The authors address the challenge that time series data can have very different patterns, like smooth changes, sudden shifts, or repeating cycles, which are hard for one single model to handle well. They created GatedLinear, a simple system that uses three different specialized parts, each good at one type of pattern, and a smart gate to decide how much to use each part for every prediction point. This approach makes the model more flexible and interpretable while being smaller and faster than many recent complex models. Tests show it works as well or better than those complicated methods on common forecasting tasks.
time series forecastingnonstationary drifttrend-seasonal basisphase-aligned recurrencelinear basesadaptive routingTri-Factorized Fusion Gateself-attentionspectral filteringmodel interpretability
Authors
Qitai Tan, Ruiwen Gu, Yilin Su, Mo Li, Xu Lin, Xiao-Ping Zhang
Abstract
Time series forecasting requires models to capture diverse, often mutually exclusive, temporal dynamics, from smooth trend continuation to nonstationary drift and strict phase-aligned recurrence. While recent deep learning models have improved accuracy, they typically force these diverse patterns through a single computational backbone governed by fixed algorithmic inductive biases (e.g., self-attention or spectral filtering). This single-mechanism approach often struggles with the profound heterogeneity of real-world series, where different variables and forecast horizons necessitate fundamentally different predictive treatments. To address this, we propose GatedLinear: a lightweight framework that frames forecasting as the adaptive routing of complementary linear bases. GatedLinear leverages a pool of three specialized mechanisms: a global trend-seasonal basis for smooth projection, a difference-based incremental basis for nonstationary drift, and a phase-aligned recurrence basis for explicit cyclic reuse. To dynamically orchestrate these distinct behaviors, we introduce a Tri-Factorized Fusion Gate that disentangles routing decisions into channel-specific preferences, horizon-aware offsets, and phase-indexed biases derived from known future time marks. This design allows the model to perform highly granular, point-wise soft routing across different predictive regimes without stacking computationally heavy neural modules. Experiments on standard benchmarks show that our method achieves state-of-the-art or highly competitive accuracy against recent complex foundational models, while offering explicitly interpretable routing patterns and operating with a substantially smaller parameter footprint.