Information Technology
System for Automatic Error Estimate Correction for a Machine Learning Model
WARF: P200120US02
Inventors: Dane Morgan, Ryan Jacobs, Glenn Palmer
The Invention
UW-Madison researchers have developed a computational method for correcting the error estimate of a prediction. Implemented in computer software, the method leverages existing data sets to compute curve descriptive values that are used as a comparator, which can ultimately be used to improve a model’s domain.
Operationally, an input dataset is split into a training dataset and a validation dataset, a predictive model and a domain model are trained, the trained predictive model and the trained domain model are validated, a predictive error value, a residual value, and a domain error value are computed, and each value is stored in output data. A domain threshold value is computed from the stored domain error values. Each predictive error value and each residual value stored in the output data is stored in in-domain output data when a respective domain error value is less than or equal to the computed domain threshold value. Curve descriptive values are computed to describe a relationship between the residual values as a function of the prediction error values stored in the in-domain output data.
Operationally, an input dataset is split into a training dataset and a validation dataset, a predictive model and a domain model are trained, the trained predictive model and the trained domain model are validated, a predictive error value, a residual value, and a domain error value are computed, and each value is stored in output data. A domain threshold value is computed from the stored domain error values. Each predictive error value and each residual value stored in the output data is stored in in-domain output data when a respective domain error value is less than or equal to the computed domain threshold value. Curve descriptive values are computed to describe a relationship between the residual values as a function of the prediction error values stored in the in-domain output data.
Key Benefits
- Uses both distances in feature space (X) and model uncertainties in target space (Y) to assess model domain
Additional Information
For More Information About the Inventors
For current licensing status, please contact Michael Carey at [javascript protected email address] or 608-960-9867