Context-Adaptive Thresholding for Conditionally Representative Monitoring and Classification

Ansgar Steland

Commonly, classifiers and monitoring procedures are trained from labeled data by optimizing an objective such as the misclassification rate. This may lead to unrepresentative condi…