首页|Researchers Submit Patent Application, 'Visualization Methods And Systems For Ev aluating Performance Of A Classifier And Selection Of Optimal Working Thresholds ', for Approval (USPTO 20240393932)
Researchers Submit Patent Application, 'Visualization Methods And Systems For Ev aluating Performance Of A Classifier And Selection Of Optimal Working Thresholds ', for Approval (USPTO 20240393932)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
News editors obtained the following quote from the background information suppli ed by the inventors:“Machine learning (ML) models have been used in healthcare such as disease diagnosis and detection. Insome cases, the predictive models ma y return a prediction, as well as a confidence score indicating themodel’s conf idence in the prediction (e.g., the probability returned by logistic regression) . A confidencescore is a number between 0 and 1 that represents the likelihood that the output of a Machine Learningmodel is correct and will satisfy a user’s request (the higher number the more likely the result of the modelmatching the user’s request). However, models may not always produce the correct confidence score (e.g.,a prediction of a class with confidence p is not correct 100*p perc ent of the time.). For example, amis-calibrated model (due to insufficient trai ning datasets and/or imbalanced training data) may result inconfidence scores d o not correspond to the probability of an answer being correct. There are method s forcalibrating machine learning models (e.g., sigmoid method, isotonic regres sion, Platt scaling, etc.) whichrequire finding a monotonic function mapping th e confidence score to correctness such as by comparingconfidence and accuracy o n the test sample.”
NETL
NSTL
