Table 12 Feature selection ablation study on both datasets.

From: Few-shot android malware classification with quantum-enhanced prototypical learning and drift detection

Dataset

Feature Set

Features

Accuracy

Reduction

CCCS-CIC-AndMal-2020

All Original

9,503

99.75%

0%

CatBoost Top-100

100

99.72%

98.95%

CatBoost Top-51

51

99.70%

99.46%

Random Selection

51

93.20%

99.46%

KronoDroid

All Original

489

99.41%

0%

CatBoost Top-50

50

99.38%

89.77%

CatBoost Top-29

29

99.33%

94.07%

Random Selection

29

94.70%

94.07%

Back to article page