Fig. 6.

Novelty scores and threshold values of KNFST, KDE and SVDD classifiers using different training dataset sizes in the one-class novelty detection scenario (applied to tyrosine, which has two instances as shown on the supplementary material). The red, green and blue crosses resemble the unknown test data, known test data and known training data, respectively. Subfigures (a) to (c) correspond to the variations of the output of the classifiers when using (a) 2.5%, (b) 10% and (c) 100% of the training dataset. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)