AI Paper Atlas

Research map / Time series / Long-term forecasting

PMDformer: Patch-Mean Decoupling Transformer for Long-term Forecasting

Ao Hu, Liangjian Wen, Jiang Duan, Yong Dai, YAN HE, Dongkai Wang, Jun Wang, Yukun Zhang, Ruoxi Jiang, Zenglin Xu

ICLR 2026 · Time series › Long-term forecasting · approach: Transformers for long-term forecasting

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
use patch-mean decoupling instead of mean-variance patch normalization (ablation)PMDformer · TrafficMSE: 0.396 -> 0.378Improved
use patch-mean decoupling instead of SAN (ablation)PMDformer · ETTh2MSE: 0.360 -> 0.337Improved
add patch-mean decoupling versus no decoupling (ablation)PMDformer · SolarMSE: 0.199 -> 0.181Improved
use patch-mean decoupling instead of mean-variance normalization (ablation)PMDformer · ETTm1 and WeatherMSE: ETTm1 0.347 -> 0.346; Weather 0.218 -> 0.217No clear effect
restrict cross-variable attention to the last patch (PVA) (ablation)PMDformer · SolarMSE: 0.186 -> 0.181Improved
add PVA to model without PVA (ablation)PMDformer · TrafficMSE: 0.386 -> 0.378Improved
use trend restoration attention instead of standard self-attention (ablation)PMDformer · TrafficMSE: 0.388 -> 0.378Improved
add TRA and PVA together (ablation)PMDformer · TrafficMSE: 0.426 -> 0.378Improved
apply PVA before TRA instead of TRA before PVA (ablation)PMDformer · SolarMSE: 0.188 (swapped) -> 0.181Improved
use cross-variable attention over k nearest patches larger than 1 (ablation)PMDformer · ETTh1 and SolarMSEHurt
use moderate patch sizes 24 to 72 instead of very small or large (ablation)PMDformer · ETTh2 and ECLMSEImproved
use PMDformer instead of TimeBaseTimeBase · eight LTSF datasetsMSE: 5.68% average MSE reductionImproved

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.

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

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