Sage is an integrated software tool for training surrogate models, assessing surrogate model quality, and refining surrogate models through intelligent adaptive sampling processes. Sage provides the following key features:
- Data Fitting Algorithms: The framework includes algorithms for fitting available data to form surrogate models. It ensures that the model captures essential features while remaining computationally tractable.
- Sampling Techniques: The framework provides techniques for intelligently selecting samples over the input space. Key adaptive sampling techniques leverage surrogate models to focus samples where they most improve model accuracy.
- Model Validation: The framework provides techniques for assessing the accuracy of the model.
- Uncertainty Quantification: The trained models provide estimates of their confidence in their predictions.