The Role of Domain-Specific Features in Malware Detection: A macOS Case Study CCS: Security and privacy Malware and its mitigation CCS: Computing methodologies Machine learning Conference: ACM Asia Conference on Computer and Communications Security; June 1–5, 2026; Bangalore, India ACM Asia Conference on Computer and Communications Security (ASIA CCS ’26), June 1–5, 2026, Bangalore, India DOI: 10.1145/3779208.3785392 ISBN: 979-8-4007-2356-8/26/06
Biagio Montaruli, Andrea Oliveri, Savino Dambra, Davide Balzarotti · 2026
Read The Role of Domain-Specific Features in Malware Detection: A macOS Case Study CCS: Security and privacy Malware and its mitigation CCS: Computing methodologies Machine learning Conference: ACM Asia Conference on Computer and Communications Security; June 1–5, 2026; Bangalore, India ACM Asia Conference on Computer and Communications Security (ASIA CCS ’26), June 1–5, 2026, Bangalore, India DOI: 10.1145/3779208.3785392 ISBN: 979-8-4007-2356-8/26/06 by Biagio Montaruli, Andrea Oliveri, Savino Dambra, Davide Balzarotti (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-03218.txt.
Description
Montaruli, Oliveri, Dambra, and Balzarotti show that macOS-specific features sharpen malware detection. Most malware research targets Windows and Android, so macOS lags. The Mach-O file format and the system itself have traits that can aid classification a great deal. The team builds new static features from those traits, such as embedded certificates, entitlements, persistence tricks, and key system APIs.
A new dataset holds 41,129 samples, with 11,413 benign and 29,716 malicious files. A detector trained with the new features hit about 98.5 percent detection. It beat all past approaches, with a mean gain of 16 percent in detection rate. A feature study showed the detector leans on the new macOS traits. A fresh set of 9,000 new files tested drift over time. The detector held at about 99.5 percent and beat the prior best by half. Dropping the new features cost about 16 points, which proves their worth. The dataset is released to the research community.
Catalog record
| CREATOR | Biagio Montaruli, Andrea Oliveri, Savino Dambra, Davide Balzarotti |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 74 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.03218