Routing Channel-Patch Dependencies in Time Series Forecasting with Graph Spectral Decomposition
Not on the map. Why?
Abstract (arXiv)
Time series forecasting has attracted significant attention in the field of AI. Previous works have revealed that the Channel-Independent (CI) strategy improves forecasting performance by modeling each channel individually, but it often suffers from poor generalization and overlooks meaningful inter-channel interactions. Conversely, Channel-Dependent (CD) strategies aggregate all channels, which may introduce irrelevant information and lead to oversmoothing. Despite recent progress, few existing methods offer the flexibility to adaptively balance CI and CD strategies in response to varying channel dependencies. To address this, we propose a generic plugin xCPD, that can adaptively model the channel-patch dependencies from the perspective of graph spectral decomposition. Specifically, xCPD first projects multivariate signals into the frequency domain using a shared graph Fourier basis, and groups patches into low-, mid-, and high-frequency bands based on their spectral energy responses. xCPD then applies a channel-adaptive routing mechanism that dynamically adjusts the degree of inter-channel interaction for each patch, enabling selective activation of frequency-specific experts. This facilitates fine-grained input-aware modeling of smooth trends, local fluctuations, and abrupt transitions. xCPD can be seamlessly integrated on top of existing CI and CD forecasting models, consistently enhancing both accuracy and generalization across benchmarks. The code is available https://github.com/Clearloveyuan/xCPD.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 39 of its 54 references that are in the map. First posted on arXiv 14 Mar 2026.
Builds on
39 of 54 references.
References · 39
- 2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,509
- 2020Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecastingcited by 2,169
- 2022A Time Series is Worth 64 Words: Long-term Forecasting with Transformerscited by 2,039
- 2021Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecastingcited by 1,676
- 2022Are Transformers Effective for Time Series Forecasting?cited by 1,424
- 2022FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecastingcited by 1,210
- 2022TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysiscited by 1,058
- 2023iTransformer: Inverted Transformers Are Effective for Time Series Forecastingcited by 931
- 2017Modeling Long- and Short-Term Temporal Patterns with Deep Neural Networkscited by 879
- 2023Time-LLM: Time Series Forecasting by Reprogramming Large Language Modelscited by 700
- 2024Unified Training of Universal Time Series Forecasting Transformerscited by 614
- 2022Non-stationary Transformers: Exploring the Stationarity in Time Series Forecastingcited by 391
27 more
- 2024TimeMixer: Decomposable Multiscale Mixing for Time Series Forecastingcited by 346
- 2021SCINet: Time Series Modeling and Forecasting with Sample Convolution and Interactioncited by 338
- 2023Long-term Forecasting with TiDE: Time-series Dense Encodercited by 332
- 2021Spectral Temporal Graph Neural Network for Multivariate Time-series Forecastingcited by 299
- 2023TSMixer: An All-MLP Architecture for Time Series Forecastingcited by 240
- 2019Recurrent Neural Networks for Time Series Forecasting: Current Status and Future Directionscited by 172
- 2023Frequency-domain MLPs are More Effective Learners in Time Series Forecastingcited by 172
- 2022FiLM: Frequency improved Legendre Memory Model for Long-term Time Series Forecastingcited by 147
- 2023FourierGNN: Rethinking Multivariate Time Series Forecasting from a Pure Graph Perspectivecited by 127
- 2024DUET: Dual Clustering Enhanced Multivariate Time Series Forecastingcited by 107
- 2023The Capacity and Robustness Trade-off: Revisiting the Channel Independent Strategy for Multivariate Time Series Forecastingcited by 101
- 2024UniTS: A Unified Multi-Task Time Series Modelcited by 94
- 2025TimeFilter: Patch-Specific Spatial-Temporal Graph Filtration for Time Series Forecastingcited by 64
- 2023MSGNet: Learning Multi-Scale Inter-Series Correlations for Multivariate Time Series Forecastingcited by 50
- 2024FilterNet: Harnessing Frequency Filters for Time Series Forecastingcited by 47
- 2024Revitalizing Multivariate Time Series Forecasting: Learnable Decomposition with Inter-Series Dependencies and Intra-Series Variations Modelingcited by 44
- 2024TimeBridge: Non-Stationarity Matters for Long-term Time Series Forecastingcited by 40
- 2024From Similarity to Superiority: Channel Clustering for Time Series Forecastingcited by 37
- 2024Adaptive Multi-Scale Decomposition Framework for Time Series Forecastingcited by 34
- 2024Rethinking Channel Dependence for Multivariate Time Series Forecasting: Learning from Leading Indicatorscited by 32
- 2023Fully-Connected Spatial-Temporal Graph for Multivariate Time-Series Datacited by 23
- 2021NeuralProphet: Explainable Forecasting at Scalecited by 22
- 2024Ada-MSHyper: Adaptive Multi-Scale Hypergraph Transformer for Time Series Forecastingcited by 21
- 2023Multi-scale Transformer Pyramid Networks for Multivariate Time Series Forecastingcited by 9
- 2025ReFocus: Reinforcing Mid-Frequency and Key-Frequency Modeling for Multivariate Time Series Forecastingcited by 6
- 2024Efficient High-Resolution Time Series Classification via Attention Kronecker Decompositioncited by 5
- 2022BASiS: Batch Aligned Spectral Embedding Spacecited by 4
Built on by
Citing papers per year: 2026: 2.
2 papers.