CATS-Linear: Classification Auxiliary Linear Model for Time Series Forecasting

2025-10-09

AI Summary

Heterogeneous time series instances require distinct linear mappings for accurate forecasting, yet existing fixed-parameter models lack adaptability. Method: This paper proposes CACI (Classification-Assisted Channel-Independent), a novel framework comprising three key components: (i) trend-seasonal decomposition to disentangle dynamic and periodic components; (ii) complex-valued linear projections to enhance representational capacity; and (iii) a supervised channel classifier enabling instance-aware dynamic routing—assigning each input to an optimal, instance-specific linear mapping path. Contribution/Results: We theoretically analyze how channel configuration affects expected risk. Empirically, under identical hyperparameters, CACI achieves state-of-the-art performance comparable to tuned baselines—significantly outperforming other fixed-parameter models across diverse benchmarks. This demonstrates CACI’s superior efficiency, generalization capability, and modeling robustness in heterogeneous time series forecasting.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Recent research demonstrates that linear models achieve forecasting performance competitive with complex architectures, yet methodologies for enhancing linear models remain underexplored. Motivated by the hypothesis that distinct time series instances may follow heterogeneous linear mappings, we propose the Classification Auxiliary Trend-Seasonal Decoupling Linear Model CATS-Linear, employing Classification Auxiliary Channel-Independence (CACI). CACI dynamically routes instances to dedicated predictors via classification, enabling supervised channel design. We further analyze the theoretical expected risks of different channel settings. Additionally, we redesign the trend-seasonal decomposition architecture by adding a decoupling -- linear mapping -- recoupling framework for trend components and complex-domain linear projections for seasonal components. Extensive experiments validate that CATS-Linear with fixed hyperparameters achieves state-of-the-art accuracy comparable to hyperparameter-tuned baselines while delivering SOTA accuracy against fixed-hyperparameter counterparts.

Problem

• Enhancing linear models for time series forecasting with auxiliary classification
• Dynamically routing instances to specialized predictors via classification
• Redesigning trend-seasonal decomposition with improved linear mapping architecture

Innovation

• Dynamic instance routing via classification auxiliary
• Decoupling-recoupling framework for trend components
• Complex-domain linear projections for seasonal components