Research map / Time series / Causal and generative time series
FracTS: Hierarchical and Autoregressive Time Series Generation
Jiayu Li, Umair Afzal, Zilong Zhao, Milad Abdollahzadeh, Uzair Javaid, Biplab Sikdar
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes FracTS, a hierarchical autoregressive time series generation method; the abstract is truncated so details beyond the motivation are unknown.
Problem
Time series generation: early autoregressive generators suffer from low sample diversity, error accumulation, and difficulty modeling long-range dependencies.
Approach
Introduces FracTS, a hierarchical and autoregressive generative approach for time series; the specific mechanism is not stated in the available abstract text.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use FracTS hierarchical autoregressive generation instead of best baseline for unconditional generation | FracTS · Unconditional (3 datasets avg) | PM W=1000: 4858 (COSCI-GAN) -> 0.010 | Improved |
| Use FracTS instead of TabularARGN for conditional generation, MD raw | FracTS · Conditional (Walmart+Simulated avg) | MD raw: 0.154 -> 0.172 | Hurt |
| Use FracTS instead of TabularARGN for conditional generation, CP agg.+cond. | FracTS · Conditional (Walmart+Simulated avg) | CP agg.+cond.: 0.345 -> 0.409 | Hurt |
| Use FracTS instead of best baseline for conditional generation, MD length | FracTS · Conditional (Walmart+Simulated avg) | MD length: 0.278 -> 0.139 | Improved |
| Use FracTS for ground-truth fidelity on simulated formulas | FracTS · Simulated dataset | GT MAE: 0.884 -> 0.545 | Improved |
| Use FracTS for training cost | FracTS · Weather | training time (s): 1560 -> 4068 | Hurt |
| Use FracTS for inference on long Walmart sequences vs TabularARGN | FracTS · Walmart | inference time (s): 300 -> 241 | Improved |
| Use FracTS for inference on Weather vs PaD-TS | FracTS · Weather | inference time (s): 10 -> 240 | Hurt |
Limitations
- Full text unavailable; only the truncated abstract was provided.
Closest papers by meaning
- Time Series Generation Under Data Scarcity: A Unified Generative Modeling Approach (NeurIPS 2025 · similarity 0.90)
- Utilizing Image Transforms and Diffusion Models for Generative Modeling of Short and Long Time Series (NeurIPS 2024 · similarity 0.89)
- TimeTok: Granularity-Controllable Time-Series Generation via Hierarchical Tokenization (NeurIPS 2026 · similarity 0.88)
- A Diffusion Model for Regular Time Series Generation from Irregular Data with Completion and Masking (NeurIPS 2025 · similarity 0.87)
- HDT: Hierarchical Discrete Transformer for Multivariate Time Series Forecasting (AAAI 2025 · similarity 0.86)
- TSGM: A Flexible Framework for Generative Modeling of Synthetic Time Series (NeurIPS 2024 · similarity 0.86)
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.
