Research map / Time series / Foundation models for time series
Rating Quality of Diverse Time Series Data by Meta-learning from LLM Judgment
Shunyu Wu, Dan Li, Wenjie Feng, Haozheng Ye, Jian Lou, See-Kiong Ng
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 meta-learning across datasets to the rater instead of training on another dataset (ablation) | TSRater · Electricity (Linear) | RMSE: 1.556 -> 1.390 | Improved |
| meta-train across datasets instead of training only on the target dataset (ablation) | TSRater · Electricity (Linear) | RMSE: 1.471 -> 1.390 | Improved |
| meta-train rather than training on Electricity (CNN) (ablation) | TSRater · Electricity (CNN) | RMSE: 1.497 -> 1.511 | No clear effect |
| fuse four criteria instead of trend only (ablation) | TSRating · Electricity (TimeMixer) | RMSE: 0.366 -> 0.318 | Improved |
| fuse four criteria instead of amplitude only (ablation) | TSRating · Weather (TimeMixer) | RMSE: 0.418 -> 0.433 | Hurt |
| fuse four criteria instead of amplitude only (ablation) | TSRating · Traffic (TimeMixer) | RMSE: 0.320 -> 0.285 | Improved |
| fuse four criteria instead of pattern only (ablation) | TSRating · Traffic (TimeMixer) | RMSE: 0.265 -> 0.285 | Hurt |
| swap TSFM encoder MOMENT for Chronos (ablation) | TSRater · Weather (TimeMixer) | RMSE: 0.433 -> 0.428 | No clear effect |
| swap LLM judge GPT-4o-mini for Gemini-2.0 (ablation) | TSRating · CBF (Informer) | accuracy: 0.648 -> 0.697 | No clear effect |
| apply TSRating selection instead of random selection | Linear · Traffic | RMSE: 0.979 -> 0.683 | Improved |
| apply TSRating selection instead of random selection | CNN · M4-Yearly | MAPE: 2.332 -> 1.782 | Improved |
| apply TSRating selection instead of random selection | Linear · CBF | accuracy: 0.294 -> 0.361 | Improved |
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.
- TimeInf — TimeInf: Time Series Data Contribution via Influence Functions (ICLR 2025) · better on 1 of 1 shared result
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
- TSQAgent: Rating Time Series Data Quality via Dedicated Agentic Reasoning (NeurIPS 2026 · similarity 0.91)
- Time-IMM: A Dataset and Benchmark for Irregular Multimodal Multivariate Time Series (NeurIPS 2025 · similarity 0.87)
- Time-MMD: Multi-Domain Multimodal Dataset for Time Series Analysis (NeurIPS 2024 · similarity 0.87)
- TSGM: A Flexible Framework for Generative Modeling of Synthetic Time Series (NeurIPS 2024 · similarity 0.86)
- Can LLMs Understand Time Series Anomalies? (ICLR 2025 · similarity 0.85)
- fev-bench: A Realistic Benchmark for Time Series Forecasting (NeurIPS 2026 · similarity 0.85)
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.
