DETECTION DEVICE, CORRECTION SYSTEM, DETECTION METHOD AND PROGRAM

摘要

In the present invention, training-data error candidates are detected with ease and certainty. Provided is a detection device that detects errors in training data for generating a classifier, and that is provided with: a first test unit that tests a classifier generated by machine learning of training data; a deletion unit that generates reduced training data in which at least one set of input data was deleted from the training data; a second test unit that tests a classifier for testing generated by machine learning of the reduced training data; and an output unit that outputs, as an error candidate, the at least one set of input data deleted from the training data in order to generate the reduced training data, in a case where there is input data that failed the test of the classifier and passed the test of the classifier for testing.