Sponsor: U.S. Geological Survey

Developers: Eileen P. Poeter, Mary C. Hill, Dan Lu, and Steffen Mehl

Webpage: https://igwmc.mines.edu/ucode-2/

Description: UCODE is one of a set of inverse modeling codes supported by the U.S. Geological Survey.
UCODE was developed for models in which the number of parameters is less than the number of observations. It can be used with existing process models to perform sensitivity analysis, data needs assessment, model calibration, prediction and uncertainty quantification.

Sponsor: Oak Ridge National Laboratory

Developers: Siyan Liu, Dan Lu, and Guannan Zhang

Webpage: https://github.com/liusiyan/PI3NN

Description: PI3NN was developed for uncertainty quantification of machine learning model predictions.
PI3NN calculates prediction intervals using three neural networks. It can precisely quantify the prediction uncertainty of the in-distribution data with a desired confidence level and accurately identify the out-of-distribution samples to avoid overconfident prediction. PI3NN is computationally efficient involving three networks training; it is reliable in training and robust in prediction without introducing extra hyperparameters; it is generalizable to various network structures and applicable to different data with no distributional assumptions.