Assisting in Auditing of Buffer Overflow Vulnerabilities via Machine Learning

Table 11

Performances of our five classifiers on Poppler 0.10.6.

Classifier	TP	FN	FP	TN	Recall (%)	TNR (%)	Precision (%)	(%)	SVFs	Sink functions	All functions
KNN	7	4	37	648	63.6	94.6	15.9	25.4	44	685	4876
DT	8	3	44	641	72.7	93.6	15.4	25.4	52
NB	10	1	52	633	90.9	92.4	16.1	27.4	62
AdaBoost	9	2	39	646	81.8	94.3	18.8	30.6	48
SVM	10	1	35	650	90.9	94.9	22.2	35.7	45