Research map / Time series / Long-term forecasting
MDMLP-EIA: Multi-domain Dynamic MLPs with Energy Invariant Attention for Time Series Forecasting
Hu Zhang, Zhien Dai, Zhaohui Tang, Yongfang Xie
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 weak seasonal MLP with AZCF fusion (ablation) | MDMLP-EIA · ETT/Weather avg | MSE: 0.339 -> 0.338 | Improved |
| use zero initialization instead of random (ablation) | AZCF · ETT/Weather avg | MSE: 0.341 -> 0.338 | No clear effect |
| use single-parameter fusion instead of two weights (ablation) | AZCF · ETT/Weather avg | MSE: 0.340 -> 0.338 | No clear effect |
| add channel-dimension fusion (ablation) | AZCF · ETTh2 | MSE: 0.374 -> 0.362 | Improved |
| add energy invariant attention vs plain addition (ablation) | MDMLP-EIA · Avg six datasets | MSE: 0.328 -> 0.321 | Improved |
| use EIA instead of MLP attention fusion (ablation) | MDMLP-EIA · Avg six datasets | MSE: 0.341 -> 0.321 | Improved |
| use EIA instead of adaptive gating (no 2x factor) (ablation) | MDMLP-EIA · Avg six datasets | MSE: 0.327 -> 0.321 | Improved |
| add EIA fusion (ablation) | Amplifier · Avg six datasets | MSE: 0.334 -> 0.328 | Improved |
| add EIA fusion (ablation) | xPatch · Avg six datasets | MSE: 0.330 -> 0.328 | Improved |
| use dynamic capacity adjustment instead of fixed 1024 neurons (ablation) | MDMLP-EIA · Avg ETTh1/Solar/Traffic/Weather | MSE: 0.354 -> 0.349 | Improved |
| use fixed 32 neurons instead of DCA on Traffic (ablation) | MDMLP-EIA · Traffic | MSE: 0.481 -> 0.610 | Hurt |
| scale hidden size with ETTh1 (ablation) | MDMLP-EIA · ETTh1 | MSE: 0.438 -> 0.436 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- xPatch — xPatch: Dual-Stream Time Series Forecasting with Exponential Seasonal-Trend Decomposition (AAAI 2025) · better on 8 of 8 shared results
- Amplifier — Amplifier: Bringing Attention to Neglected Low-Energy Components in Time Series Forecasting (AAAI 2025) · better on 3 of 3 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
- Unlocking the Power of Patch: Patch-Based MLP for Long-Term Time Series Forecasting (AAAI 2025 · similarity 0.86)
- Introducing Spectral Attention for Long-Range Dependency in Time Series Forecasting (NeurIPS 2024 · similarity 0.86)
- Revisiting Attention for Multivariate Time Series Forecasting (AAAI 2025 · similarity 0.85)
- Adaptive Multi-Scale Decomposition Framework for Time Series Forecasting (AAAI 2025 · similarity 0.85)
- CATS: Enhancing Multivariate Time Series Forecasting by Constructing Auxiliary Time Series as Exogenous Variables (ICML 2024 · similarity 0.85)
- SOFTS: Efficient Multivariate Time Series Forecasting with Series-Core Fusion (NeurIPS 2024 · similarity 0.84)
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.
