microsoft / TimeCraft

What it does

• Multi-Domain Time Series Generation. Robust cross-domain generalization enabled by few-shot learning, requiring minimal data from new domains.
• Controllable Generation. Natural language text-based control allows users to specify desired characteristics like trends or seasonality.
• Target-Aware Generation. Synthesized data is explicitly optimized to improve downstream model performance on tasks like forecasting or classification.
• Diffusion-Based Framework. Ensures high-fidelity, stable, and diverse time series through powerful diffusion modeling.
• Automated Time Series Description. Generates descriptive text to enhance interpretability and support paired training or analysis.