Patent · US Active

Training, validating, and monitoring artificial intelligence and machine learning models

US10990901B2 · kind B2 · utility

8Cited by
0References
20Claims
0Family size

Assignee

Inventors

Key dates

Filing dateAug 31, 2018
Grant dateApr 27, 2021
Priority date
Expiry dateAug 31, 2038

Classification

  • Technology area (CPC G)Physics
  • CPC primaryG06N20/10
  • WIPO fieldComputer technology
  • WIPO sectorElectrical engineering

Abstract

A device identifies training data and scoring data for a model, and removes bias from the training data to generate unbiased training data. The device trains the model with the unbiased training data to generate trained models, and processes the trained models, with the scoring data, to generate scores for the trained models. The device selects a trained model, from the trained models, based on model metrics and the scores, and processes a training sample, with the trained model, to generate first results, wherein the training sample is created based on the unbiased training data and production data. The device processes a production sample, with the trained model, to generate second results, wherein the production sample is created based on the production data and the training sample. The device provides the trained model for use in a production environment based on the first results and the second results.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.