New OATS strategy enhances Time Series Foundation Models with dynamic data augmentation

Researchers have developed OATS, a novel online data augmentation strategy for Time Series Foundation Models (TSFMs). This method dynamically generates synthetic data tailored to specific training stages, using valuable training samples as guiding signals. OATS employs a diffusion-based framework for realistic time series generation and an explore-exploit mechanism to balance efficiency and effectiveness. Experiments show OATS significantly outperforms standard training and existing static data augmentation techniques across multiple datasets and TSFM architectures. AI

IMPACT Enhances the training of time series models, potentially improving performance in forecasting and anomaly detection tasks.

RANK_REASON The item is an academic paper detailing a new method for time series foundation models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• IArxiv Recommender
• Junwei Deng
• ScienceCast
• Time Series Foundation Models
• TSFMs

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →