Research map / Time series / Foundation models for time series
One Batch Is Enough: A Unified Dataset Condensation Framework for General Time Series Analysis
Wei Shao, Ziquan Fang, Zheqi Lu, Yongfeng Su, Yuzhu Wang, Yunjun Gao
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 |
|---|---|---|---|
| replace task-specific condensation with unified tri-space alignment (UniTSC) for forecasting | UniTSC · Forecasting (7 datasets, 96 samples) | MSE: average MSE reduction of 7% relative to the second-best method | Improved |
| use UniTSC condensation instead of baselines for anomaly detection | UniTSC · Anomaly detection (5 datasets) | F1-Score: 80.75 -> 81.63 (vs CondTSC) | Improved |
| condense imputation data with UniTSC instead of forecasting-oriented CondTSF | UniTSC · ETTh1 imputation | MSE: 0.236 -> 0.205 | Improved |
| use UniTSC synthetic data (5 per class) instead of coreset selection for classification | UniTSC · UEA classification | average accuracy: up to 21.3% average accuracy improvement over coreset methods | Improved |
| condensed data trained on different architectures (DLinear, TimesNet, PatchTST) | UniTSC · ETTh1 forecasting cross-architecture | MSE | Improved |
| remove individual modules (multi-view encoder, frequency or topology alignment) (ablation) | UniTSC · ablation (appendix, not shown in the extract) | performance | Improved |
| condensed set stays at minimal budget for classification with high intra-class variance | UniTSC · UEA classification | accuracy | Hurt |
Closest papers by meaning
- Time Series Generation Under Data Scarcity: A Unified Generative Modeling Approach (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)
- TimeTok: Granularity-Controllable Time-Series Generation via Hierarchical Tokenization (NeurIPS 2026 · similarity 0.87)
- UniTS: A Unified Multi-Task Time Series Model (NeurIPS 2024 · similarity 0.87)
- TimePoint: Accelerated Time Series Alignment via Self-Supervised Keypoint and Descriptor Learning (ICML 2025 · similarity 0.86)
- TimeBase: The Power of Minimalism in Efficient Long-term Time Series Forecasting (ICML 2025 · 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.
