Skip to main content
Cybersecurity

Table 10 PUMD and multiple comparison algorithms analysis

From: PUMD: a PU learning-based malicious domain detection framework

Method

Strategy

Base classifier

Labeled sample

Full SampleSet training

Sample weight

Customized sample weight

Mali DN

Benign DN

PUMD

Two-step

RF

\(\checkmark\)

–

\(\checkmark\)

\(\checkmark\)

\(\checkmark\)

PU_biased

Biased Learning

\(\checkmark\)

–

\(\checkmark\)

\(\checkmark\)

–

PU_empirical

Incorporation with Class Priori

\(\checkmark\)

–

\(\checkmark\)

\(\checkmark\)

–

mix_supervised*

Supervised

\(\checkmark\)

\(\checkmark\)(mix)

–

–

–

pure_supervised

\(\checkmark\)

\(\checkmark\)(pure)

–

–

–

unsupervised

Unsupervised

iForest

–

–

\(\checkmark\)

–

–

  1. *baseline
Back to article page