Research map / Time series / Long-term forecasting
Unifying Channel Independence and Mixing: Multi-Scale Patch Recursion for Global–Local Representation Synergy in Multivariate Time Series Forecasting
Wenhao Zhang, Chun Zhang, Wei Bai, Ning Zhang, Changxia Gao, Yuxin Jia, Chenhao Shi, Shaoxiong Pang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add linear fusion layer over Dual-GLF and CE outputs (ablation) | FTP · PEMS04 (avg MSE) | MSE: 0.250 -> 0.094 | Improved |
| add concatenation of original embedding (residual injection) (ablation) | FTP · Electricity (avg MSE) | MSE: 0.247 -> 0.162 | Improved |
| add GLF-CM (channel-mixed branch) (ablation) | FTP · PEMS04 (avg MSE) | MSE: 0.158 -> 0.094 | Improved |
| add GLF-CI (channel-independent branch) (ablation) | FTP · Solar (avg MSE) | MSE: 0.240 -> 0.223 | Improved |
| add Channel Enhancement (CE) module (ablation) | FTP · PEMS04 (avg MSE) | MSE: 0.122 -> 0.094 | Improved |
| fuse CI and CM views (Dual-GLF) instead of either single view (ablation) | FTP · ECL, ETTh2, ETTm1, PEMS08 | MSE | Mixed |
| fuse Global and Local branches instead of one branch (level=1) (ablation) | FTP · ETTh2, ETTm2, PEMS07, Weather | MSE | Improved |
| increase recursion level from 1 to 14 (ablation) | FTP · Electricity, ETTh1, ETTh2, PEMS03 | MSE | Improved |
| use multi-scale patch spans instead of single scale (level 8) (ablation) | FTP · ETT datasets, long horizon T=1024 | MSE | Improved |
| enable deeper recursion with single scale (SS8 vs SS1) (ablation) | FTP · ETT datasets | MSE | Improved |
| lengthen look-back window from 96 to 336 (ablation) | FTP · ETTh1, ETTh2, ETTm1, ETTm2 | MSE | Mixed |
| vary unit patch length and stride (ablation) | FTP · ETTh2, Weather, Solar, PEMS04, Electricity | MSE/MAE | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SOFTS — SOFTS: Efficient Multivariate Time Series Forecasting with Series-Core Fusion (NeurIPS 2024) · better on 5 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- SOFTS: Efficient Multivariate Time Series Forecasting with Series-Core Fusion (NeurIPS 2024 · similarity 0.91)
- Unlocking the Power of Patch: Patch-Based MLP for Long-Term Time Series Forecasting (AAAI 2025 · similarity 0.91)
- TimeCHEAT: A Channel Harmony Strategy for Irregularly Sampled Multivariate Time Series Analysis (AAAI 2025 · similarity 0.90)
- Breaking Silos: Adaptive Model Fusion Unlocks Better Time Series Forecasting (ICML 2025 · similarity 0.89)
- CSformer: Combining Channel Independence and Mixing for Robust Multivariate Time Series Forecasting (AAAI 2025 · similarity 0.89)
- TiRex-2: Generalizing TiRex to Multivariate Data and Streaming (NeurIPS 2026 · similarity 0.89)
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.
