Using Graph Representation in Host-Based Intrusion Detection

Table 2

Classification result of different embedding methods.

Embedding methods	Classifier	AUC	Recall	Accuracy
Bag-of-words	KNN	0.9172	0.8983	0.9352
TF-IDF	KNN	0.8917	0.8644	0.9172
N-gram	KNN	0.9036	0.8729	0.9327
Word2Vec	KNN	0.9079	0.8814	0.9329
GRSE	KNN	0.9207	0.8983	0.9419
Bag-of-words	MLP	0.9483	1.0	0.8993
TF-IDF	MLP	0.9495	0.9915	0.9097
N-gram	MLP	0.9445	1.0	0.8919
Word2Vec	MLP	0.9455	0.9830	0.9100
GRSE	MLP	0.9555	0.9915	0.9213