METHOD AND SYSTEM FOR BUILDING TRAINING DATABASE USING AUTOMATIC ANOMALY DETECTION AND AUTOMATIC LABELING TECHNOLOGY

摘要

The present invention relates to a method and system for building a learning database using an automatic detection of abnormal data and automatic labeling technology, wherein the method learns a labeled first training data population to classify first and second recognition models building each of them independently; generating a second labeled training data population by applying the first recognition model to an unlabeled raw data population; classifying the abnormal data by applying the second recognition model to the first training data population; and refining the abnormal data from the first training data population and updating the first training data population by adding the second training data population.