Using Machine Learning to Analyze and Detect Anomalies in SELinux Security Policies
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Analysis of Security-Enhanced Linux (SELinux) policies requires extensive manual effort to identify violations and security mis-configurations. Current tools employ mathematical abstractions that, while theoretically sound, produce outputs that practitioners struggle to interpret effectively. Automated approaches using machine learning have shown promise but fail to capture the complex relationships inherent in SELinux policies. Here we present a novel approach combining graph-based policy representation with neural networks to automate SELinux policy analysis. Our approach represents policies as graph structures and transforms these structures into meaningful vector embeddings to learn continuous feature representations that preserve policy neighborhoods and violation patterns. We develop a flexible policy analysis framework that processes these representations through Random Forest, Support Vector Machine (SVM), and Multi-Layer Perceptron (MLP) models to detect violations. Our experimental results demonstrate that this approach achieves 95% accuracy in identifying security violations while maintaining balanced precision and recall metrics, significantly outperforming existing analysis techniques. Through extensive evaluation on synthetic policy datasets derived from production systems, we show that our method effectively captures diverse violation patterns including separation of duty violations, domain transition issues, and unauthorized access paths. Overall, our work presents an efficient approach for automated, interpretable SELinux policy analysis that bridges the gap between theoretical security models and practical policy management.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.004 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it