Research map / Time series / Long-term forecasting
TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis
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 |
|---|---|---|---|
| Model temporal variation with TimesNet's multi-period 2D representation | Time series forecasting · Long-term forecasting benchmarks | MSE | Mixed |
| Use TimesNet for short-term forecasting | M4 · M4 | SMAPE: 11.851 -> 11.829 | Helped |
| Use TimesNet for imputation | Time series imputation · ETTm1, ETTm2, ETTh1, ETTh2, Electricity, Weather | MSE | Helped |
| Use TimesNet for classification | UEA classification · UEA | Accuracy: 72.5% -> 73.6% | Helped |
| Use TimesNet with ResNeXt as the 2D backbone | TimesNet · Anomaly detection (SMD, MSL, SMAP, SWaT, PSM) | Avg F1: 80.50 -> 86.34 | Helped |
| Replace ResNeXt with ResNet (ablation) | TimesNet · Anomaly detection (SMD, MSL, SMAP, SWaT, PSM) | Avg F1: 86.34 -> 84.81 | Hurt |
| Replace ResNeXt with Swin Transformer (ablation) | TimesNet · Anomaly detection (SMD, MSL, SMAP, SWaT, PSM) | Avg F1: 86.34 -> 86.11 | Hurt |
| Replace ResNeXt with ConvNeXt (ablation) | TimesNet · Anomaly detection (SMD, MSL, SMAP, SWaT, PSM) | Avg F1: 86.34 -> 86.89 | Helped |
| Use independent rather than shared Inception parameters for each period tensor (ablation) | TimesNet · Anomaly detection (SMD, MSL, SMAP, SWaT, PSM) | Avg F1: 85.49 -> 85.60 | Helped |
| Add Autoformer deep decomposition (ablation) | TimesNet · Anomaly detection (SMD, MSL, SMAP, SWaT, PSM) | Avg F1: 85.49 -> 85.29 | Hurt |
| Transform raw input instead of deep features in each TimesBlock (ablation) | TimesNet · Anomaly detection (SMD, MSL, SMAP, SWaT, PSM) | Avg F1: 85.49% -> 84.85% | Hurt |
| Sum period representations directly rather than using amplitude-based weights (ablation) | TimesNet · Anomaly detection (SMD, MSL, SMAP, SWaT, PSM) | Avg F1: 85.49 -> 84.57 | Hurt |
Closest papers by meaning
- Times2D: Multi-Period Decomposition and Derivative Mapping for General Time Series Forecasting (AAAI 2025 · similarity 0.89)
- Irregular Multivariate Time Series Forecasting: A Transformable Patching Graph Neural Networks Approach (ICML 2024 · similarity 0.87)
- TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting (ICLR 2024 · similarity 0.87)
- TimeMixer++: A General Time Series Pattern Machine for Universal Predictive Analysis (ICLR 2025 · similarity 0.87)
- TimeEmb: A Lightweight Static-Dynamic Disentanglement Framework for Time Series Forecasting (NeurIPS 2025 · similarity 0.87)
- Addressing Spatial-Temporal Heterogeneity: General Mixed Time Series Analysis via Latent Continuity Recovery and Alignment (NeurIPS 2024 · 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.