R. Barandela (Mexico, Cuba), J.S. Sánchez (Spain), and E. Rangel (Mexico)
Learning Algorithms; Machine Learning; Imperfectly Supervised; Decontamination.
Imperfectly supervised situations (machine learning applications where the assumption of label correctness does not hold for all the elements of the training sample) are very frequent in real tasks. The Decontamination methodology has recently been proposed to cope with these situations. In the present paper, two modifications of this methodology are introduced. The basic aim of these modifications is to cope with those real applications where available information can help to refine the knowledge about the kind of contamination that is potentially present in the training sample.
Important Links:
Go Back
