Research map / Time series / Foundation models for time series
Multi-Scale Hypergraph Meets LLMs: Aligning Large Language Models for Time Series Analysis
Zongjiang Shang, Dongliang Cui, Binqing Wu, Ling Chen
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 mixture of prompts (MoP) (ablation) | MSH-LLM · Traffic (horizon 96) | MSE: 0.399 -> 0.365 | Improved |
| add learnable prompts (ablation) | MSH-LLM · Traffic (horizon 96) | MSE: 0.373 -> 0.365 | Improved |
| add data-correlated prompts (ablation) | MSH-LLM · Traffic (horizon 96) | MSE: 0.368 -> 0.365 | Improved |
| add capability-enhancing prompts (ablation) | MSH-LLM · Traffic (horizon 96) | MSE: 0.375 -> 0.365 | Improved |
| use full 32-layer LLaMA-7B instead of 12 layers (ablation) | MSH-LLM · Traffic (horizon 96) | MSE: 0.370 -> 0.365 | Improved |
| use LLaMA-7B instead of 6-layer GPT-2 (ablation) | MSH-LLM · Traffic (horizon 96) | MSE: 0.393 -> 0.365 | Improved |
| multi-scale hypergraph alignment vs S2IP-LLM | MSH-LLM · Traffic long-term forecasting | MSE: 0.406 -> 0.381 | Improved |
| multi-scale hypergraph alignment vs S2IP-LLM | MSH-LLM · ETTh1 long-term forecasting | MSE: 0.405 -> 0.402 | Improved |
| multi-scale hypergraph alignment vs AutoTimes | MSH-LLM · M4 short-term forecasting | SMAPE: 11.831 -> 11.659 | Improved |
| multi-scale hypergraph alignment vs S2IP-LLM | MSH-LLM · ETT (avg) few-shot 5% | MSE: 0.445 -> 0.421 | 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.
- AutoTimes — AutoTimes: Autoregressive Time Series Forecasters via Large Language Models (NeurIPS 2024) · better on 3 of 3 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
- Context-Alignment: Activating and Enhancing LLMs Capabilities in Time Series (ICLR 2025 · similarity 0.91)
- Time-LLM: Time Series Forecasting by Reprogramming Large Language Models (ICLR 2024 · similarity 0.90)
- TimeCMA: Towards LLM-Empowered Multivariate Time Series Forecasting via Cross-Modality Alignment (AAAI 2025 · similarity 0.89)
- Position: What Can Large Language Models Tell Us about Time Series Analysis (ICML 2024 · similarity 0.89)
- STEM-LTS: Integrating Semantic-Temporal Dynamics in LLM-driven Time Series Analysis (AAAI 2025 · similarity 0.89)
- Empowering Time Series Analysis with Large-Scale Multimodal Pretraining (NeurIPS 2026 · similarity 0.88)
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.
