associating the population descriptor with the trained model

CN113366499BActive Publication Date: 2026-07-07KONINKLIJKE PHILIPS NV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
KONINKLIJKE PHILIPS NV
Filing Date
2020-01-23
Publication Date
2026-07-07

AI Technical Summary

Technical Problem

Existing technologies struggle to effectively handle records that differ significantly from the training dataset in the execution environment, leading to inaccurate model outputs, particularly in fully automated or interactive applications, which can impact the accuracy of clinical applications and user trust.

Method used

By identifying a population descriptor associated with the training dataset, the characteristic distribution of training records is characterized. This allows for the determination of whether a record is consistent with the population descriptor without accessing the intermediate output of the model, thereby generating consistency or inconsistency signals to adjust model parameters or warn the user.

Benefits of technology

It improves the accuracy and reliability of model output, avoids misuse in inconsistent situations, reduces storage and processing requirements, circumvents legal and privacy constraints, and expands the scope of model applications.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113366499B_ABST
    Figure CN113366499B_ABST
Patent Text Reader

Abstract

A model can be trained on a training dataset, e.g., for a medical image processing task or a medical signal processing task. Systems and computer-implemented methods are provided for associating a population descriptor with a trained model and using the population descriptor to determine whether a record to which the model is to be applied conforms to the population descriptor. The population descriptor characterizes a distribution of one or more characteristic features over the training dataset, where the characteristic features characterize the training records and / or characterize model outputs provided when the trained model is applied to the training records. For example, the model can only be applied to records that conform to the population descriptor, or model outputs for non-conforming records can be flagged as possibly untrustworthy.
Need to check novelty before this filing date? Find Prior Art