“…Our empirical results indicate that PayloadEmbeddings reaches 92%-99% accuracy, precision, recall, and F1-score on ISOT, UNSW-NB15, CICIDS-2017, and CIC DoS datasets, and 75%-82% accuracy, precision, recall, and F1score on the other datasets, using kNN. We compare our approach to ten other traditional and state-of-the-art techniques, including PAYL [4], McPAD [6], HMMPayl [9], AEIDS [12], HAST [11], OCPAD [7], EsPADA [22], CBID [23], PL-RNN [13], and Packet2Vec [24] over the same datasets and show that our approach performs better over most of the datasets.…”