Research map / Time series / Long-term forecasting
TSGDiff: Rethinking Synthetic Time Series Generation from a Pure Graph Perspective
Lifeng Shen, Xuyang Li, Lele Long
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 KL divergence loss (ablation) | TSGDiff · ETTh | Context-FID: 44.467 -> 0.224 | Improved |
| add KL divergence loss (ablation) | TSGDiff · ETTh | Discriminative score: 0.500 -> 0.056 | Improved |
| add KL divergence loss (ablation) | TSGDiff · ETTh | Topo-FID: 0.787 -> 0.986 | Improved |
| add denoising loss (ablation) | TSGDiff · ETTh | Discriminative score: 0.432 -> 0.056 | Improved |
| add denoising loss (ablation) | TSGDiff · ETTh | Context-FID: 3.922 -> 0.224 | Improved |
| add denoising loss (ablation) | TSGDiff · ETTh | Predictive score: 0.019 -> 0.020 | No clear effect |
| add Fourier loss (ablation) | TSGDiff · ETTh | Topo-FID: 0.883 -> 0.986 | Improved |
| add Fourier loss (ablation) | TSGDiff · ETTh | Context-FID: 0.407 -> 0.224 | Improved |
| add Fourier loss (ablation) | TSGDiff · ETTh | Discriminative score: 0.061 -> 0.056 | No clear effect |
| use TSGDiff instead of Diffusion-TS | time series generation · Six datasets | average Discriminative score: 50% improvement | Improved |
| use TSGDiff instead of Diffusion-TS | time series generation · Weather | Context-FID: 1.161 -> 0.353 | Improved |
| use TSGDiff instead of Diffusion-TS | time series generation · ETTh | Context-FID: 0.225 -> 0.224 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Diffusion-TS — Diffusion-TS: Interpretable Diffusion for General Time Series Generation (ICLR 2024) · better on 14 of 14 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
- TSGM: A Flexible Framework for Generative Modeling of Synthetic Time Series (NeurIPS 2024 · similarity 0.87)
- SigDiffusions: Score-Based Diffusion Models for Time Series via Log-Signature Embeddings (ICLR 2025 · similarity 0.86)
- CausalTime: Realistically Generated Time-series for Benchmarking of Causal Discovery (ICLR 2024 · similarity 0.86)
- Diffusion-TS: Interpretable Diffusion for General Time Series Generation (ICLR 2024 · similarity 0.86)
- TimeDART: A Diffusion Autoregressive Transformer for Self-Supervised Time Series Representation (ICML 2025 · similarity 0.86)
- TimeTok: Granularity-Controllable Time-Series Generation via Hierarchical Tokenization (NeurIPS 2026 · 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.
