Research map / Time series / Long-term forecasting
Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecasting
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 progressive decomposition architecture (ablation) | Transformer · ETTm2 predict-336 | MSE: 1.413 -> 0.375 | Helped |
| add progressive decomposition architecture (ablation) | Informer · ETTm2 predict-720 | MSE: 3.379 -> 0.822 | Helped |
| pre-decompose series and forecast parts separately (Sep) (ablation) | Transformer · ETTm2 predict-720 | MSE: 2.672 -> 3.200 | Hurt |
| pre-decompose series and forecast parts separately (Sep) (ablation) | Informer · ETTm2 predict-336 | MSE: 1.363 -> 1.469 | Hurt |
| Auto-Correlation instead of full self-attention (ablation) | Autoformer · ETTm2 input-96 predict-720 | MSE: 0.537 -> 0.422 | Helped |
| Auto-Correlation instead of ProbSparse attention (ablation) | Autoformer · ETTm2 input-96 predict-720 | MSE: 0.822 -> 0.422 | Helped |
| Auto-Correlation instead of full attention (ablation) | Autoformer · ETTm2 input-336 predict-1440 | memory | Helped |
| add past information to decoder input (I+O vs O) (ablation) | Autoformer · ETTm2 input-96 predict-336 | MSE: 0.360 -> 0.333 | Helped |
| Autoformer vs Informer | long-term forecasting · Exchange predict-336 | MSE: 1.672 -> 0.509 | Helped |
Closest papers by meaning
- SAMformer: Unlocking the Potential of Transformers in Time Series Forecasting with Sharpness-Aware Minimization and Channel-Wise Attention (ICML 2024 · similarity 0.90)
- iTransformer: Inverted Transformers Are Effective for Time Series Forecasting (ICLR 2024 · similarity 0.88)
- TimeDART: A Diffusion Autoregressive Transformer for Self-Supervised Time Series Representation (ICML 2025 · similarity 0.87)
- Timer-XL: Long-Context Transformers for Unified Time Series Forecasting (ICLR 2025 · similarity 0.87)
- Self-Supervised Contrastive Learning for Long-term Forecasting (ICLR 2024 · similarity 0.87)
- Sequence Complementor: Complementing Transformers for Time Series Forecasting with Learnable Sequences (AAAI 2025 · similarity 0.87)
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.