Static malware classifiers learn packing artifacts and dataset composition biases rather than malicious semantics, as diagnosed by TRUSTEE interpretability across controlled dataset variations.
An in-depth look into the win32 portable executable file format, part 2
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CR 1years
2026 1verdicts
UNVERDICTED 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Beyond the Wrapper: Identifying Artifact Reliance in Static Malware Classifiers using TRUSTEE
Static malware classifiers learn packing artifacts and dataset composition biases rather than malicious semantics, as diagnosed by TRUSTEE interpretability across controlled dataset variations.