AI Paper Atlas

Research map / Time series / Long-term forecasting

Routing Channel-Patch Dependencies in Time Series Forecasting with Graph Spectral Decomposition

Dongyuan Li, Shun Zheng, Chang Xu, Jiang Bian, Renhe Jiang

ICLR 2026 · Time series › Long-term forecasting · approach: Graph-based multivariate forecasting · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add xCPD plugin to backbonesTSMixer, DLinear, PatchTST, TimesNet · 9 long-term forecasting datasetsMSE/MAE: DLinear Traffic avg MSE 0.625 -> 0.606Improved
add xCPD instead of CCM or LIFTDLinear · Traffic avgMSE: CCM 0.614 / LIFT 0.620 -> 0.606Improved
use xCPD vs CCM on ETTm2PatchTST · ETTm2 avgMSE: CCM 0.306 vs xCPD 0.340Hurt
add shared Fourier basis (ablation)DLinear+xCPD · ETTh1MSE: 0.462 -> 0.445Improved
add learnable frequency partition (ablation)DLinear+xCPD · ETTh1MSE: 0.448 -> 0.445Improved
add energy-based node grouping (ablation)DLinear+xCPD · ETTh1MSE: 0.461 -> 0.445Improved
add spectral filters (ablation)DLinear+xCPD · TrafficMSE: 0.623 -> 0.606Improved
use spectral filters rather than Top-K/Random-K/RegionTop-K/TimeFilter (ablation)DLinear+xCPD · ETTh1MSE: Top-K 0.464, TimeFilter 0.457 -> 0.445Improved
add expert diversity and balance regularizers (ablation)DLinear+xCPD · TrafficMSE: 0.609 -> 0.606Improved
add xCPDDLinear · M4 averageSMAPE: 13.63 -> 12.43Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy