Research map / Time series / Long-term forecasting
Effectively Modeling Time Series with Simple Discrete State Spaces
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 |
|---|---|---|---|
| Use a companion-matrix SSM instead of S4 | Time-series SSM modeling · Synthetic AR(6) | forecasting accuracy | Helped |
| Use a companion-matrix SSM instead of S4 | SpaceTime · Informer forecasting tasks | MSE | Helped |
| Use closed-loop forecasting rather than horizon-specific direct prediction | SpaceTime · ETTh datasets, longer test horizons | MSE | Helped |
| Use SpaceTime rather than NLinear for long-horizon forecasting | ETTh datasets · Long-horizon forecasting settings | MSE | Helped |
| Use the efficient companion-matrix filter algorithm rather than naive matrix powering | SpaceTime · Companion SSM filter computation | wall-clock speed: 2× | Helped |
| Use SpaceTime rather than a parameter-matched Transformer | ETTh1 training · ETTh1, lag and horizon 720 | wall-clock training time: 73% | Helped |
| Use SpaceTime rather than a parameter-matched LSTM | ETTh1 training · ETTh1, lag and horizon 720 | wall-clock training time: 80% | Helped |
| Use SpaceTime rather than NLinear | ETTh1 · ETTh1, horizon 96 | MSE: 0.053 -> 0.054 | Hurt |
| Use SpaceTime rather than FILM | ETTm1 · ETTm1, horizon 720 | MSE: 0.071 -> 0.074 | Hurt |
| Use SpaceTime rather than S4 | PTB-XL ECG classification · PTB-XL, All | AUROC: 0.938 -> 0.936 | Hurt |
| Use SpaceTime rather than S4 | Speech Commands classification · Speech Commands | accuracy: 98.32 -> 97.29 | Hurt |
| Train to forecast fixed horizons rather than use closed-loop horizon transfer | Other direct multi-step forecasting models · Unseen longer horizons | forecasting accuracy | Hurt |
Closest papers by meaning
- From Generalization Analysis to Optimization Designs for State Space Models (ICML 2024 · similarity 0.89)
- $S^2$IP-LLM: Semantic Space Informed Prompt Learning with LLM for Time Series Forecasting (ICML 2024 · similarity 0.89)
- Chimera: Effectively Modeling Multivariate Time Series with 2-Dimensional State Space Models (NeurIPS 2024 · similarity 0.89)
- Time-LLM: Time Series Forecasting by Reprogramming Large Language Models (ICLR 2024 · similarity 0.88)
- LogSig-SSM: Time-Series Modelling with Multi-Scale Log-Signature Compression for State-Space Models (NeurIPS 2026 · similarity 0.88)
- Enhancing Foundation Models for Time Series Forecasting via Wavelet-based Tokenization (ICML 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.