Research map / Time series / Foundation models for time series
ProSAR: Prototype-Guided Semantic Augmentation and Refinement for Time Series Contrastive Learning
Caiyi Yang, Chenglin Li, Hao Zhang, Weijia Lu, ZHIFEI YANG, Wenrui Dai, xiaodong Zhang, Xiaofeng Ma, Can Zhang, Junni Zou, Hongkai Xiong
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 |
|---|---|---|---|
| use prototype-guided semantic augmentation instead of AutoTCL's augmentation | ProSAR · Univariate forecasting avg | MSE: 0.161 -> 0.151 | Improved |
| use prototype-guided semantic augmentation instead of AutoTCL's augmentation | ProSAR · Multivariate forecasting avg | MSE: 0.571 -> 0.561 | Improved |
| add DTW-based semantic segmentation (ablation) | ProSAR · Univariate forecasting | MSE: 0.172 -> 0.151 | Improved |
| add frequency (STFT) alignment (ablation) | ProSAR · Univariate forecasting | MSE: 0.178 -> 0.151 | Improved |
| add dynamic clustering of prototypes (ablation) | ProSAR · Univariate forecasting | MSE: 0.175 -> 0.151 | Improved |
| add decoder feedback (ablation) | ProSAR · Univariate forecasting | MSE: 0.164 -> 0.151 | Improved |
| add dual prototypes for input-space grounding (ablation) | ProSAR · Univariate forecasting | MSE: 0.158 -> 0.151 | Improved |
| apply ProSAR instead of FreRA in data-scarce transfer | ProSAR · SHAR transfer, 3 source domains | accuracy: 0.525 -> 0.553 | Improved |
| apply ProSAR instead of FreRA with 19 source domains | ProSAR · SHAR transfer, 19 sources, target 5 | accuracy: 0.698 -> 0.695 | No clear effect |
| apply ProSAR instead of FreRA with 1% labels | ProSAR · Semi-supervised 1% labels avg | accuracy: 0.836 -> 0.853 | Improved |
| add DTW prototype alignment to training | ProSAR · training cost | training time: 2.5% increase in total training time | 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.
- TimesURL — TimesURL: Self-Supervised Contrastive Learning for Universal Time Series Representation Learning (AAAI 2024) · 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
- Parametric Augmentation for Time Series Contrastive Learning (ICLR 2024 · similarity 0.88)
- Inference via Interpolation: Contrastive Representations Provably Enable Planning and Inference (NeurIPS 2024 · similarity 0.87)
- Time-VLM: Exploring Multimodal Vision-Language Models for Augmented Time Series Forecasting (ICML 2025 · similarity 0.86)
- Self-Supervised Contrastive Learning for Long-term Forecasting (ICLR 2024 · similarity 0.86)
- Time Series Generation Under Data Scarcity: A Unified Generative Modeling Approach (NeurIPS 2025 · similarity 0.86)
- $S^2$IP-LLM: Semantic Space Informed Prompt Learning with LLM for Time Series Forecasting (ICML 2024 · 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.
