Research map / Time series / Foundation models for time series
PATRA: Pattern-Aware Alignment and Balanced Reasoning for Time Series Question Answering
Junkai Lu, Peng Chen, Xingjian Wu, Yang Shu, Chenjuan Guo, Christian S. Jensen, Bin Yang
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 pattern-aware alignment (trend, seasonal, full) instead of none (ablation) | PATRA · TSQA Reasoning | accuracy: 28.37 -> 44.59 | Improved |
| align three patterns instead of a single pattern (ablation) | PATRA · TSQA Prescience | accuracy: 37.03 -> 52.78 | Improved |
| add RL reasoning-enhanced stage after SFT (ablation) | PATRA · TSQA Reasoning | accuracy: 13.51 -> 44.59 | Improved |
| add reward balancing to [0, 2] across tasks (ablation) | PATRA GRPO training · TSQA Comprehension | accuracy: 39.65 -> 56.03 | Improved |
| add reward balancing to [0, 2] across tasks (ablation) | PATRA GRPO training · TSQA Prescience | accuracy: 35.18 -> 52.78 | Improved |
| fine-tune ChatTS on TSQA training data | ChatTS-7B · TSQA Comprehension | accuracy: 25.00 -> 44.83 | Improved |
| fine-tune TimeMQA-Qwen-7B on TSQA training data | TimeMQA-Qwen-7B · TSQA Prescience | accuracy: 27.77 -> 26.85 | Hurt |
| fine-tune DeepSeek-R1-7B on TSQA training data | DeepSeek-R1-7B · TSQA Comprehension | accuracy: 12.93 -> 40.52 | Improved |
| PATRA multimodal deep alignment vs text-only Qwen2.5 backbone | PATRA · TSQA Recognition | accuracy: 45.51 -> 64.69 | Improved |
| latent decomposition into trend and seasonal components | PATRA · synthetic series | Pearson correlation with ground-truth trend: 0.986 | 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.
- ITFormer 7B — ITFormer: Bridging Time Series and Natural Language for Multi-Modal QA with Large-Scale Multitask Dataset (ICML 2025) · better on 2 of 2 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
- ITFormer: Bridging Time Series and Natural Language for Multi-Modal QA with Large-Scale Multitask Dataset (ICML 2025 · similarity 0.86)
- Context-Alignment: Activating and Enhancing LLMs Capabilities in Time Series (ICLR 2025 · similarity 0.84)
- Time-LLM: Time Series Forecasting by Reprogramming Large Language Models (ICLR 2024 · similarity 0.83)
- ReXTime: A Benchmark Suite for Reasoning-Across-Time in Videos (NeurIPS 2024 · similarity 0.83)
- TimeCMA: Towards LLM-Empowered Multivariate Time Series Forecasting via Cross-Modality Alignment (AAAI 2025 · similarity 0.83)
- LAST SToP for Modeling Asynchronous Time Series (ICML 2025 · similarity 0.83)
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.
