Adaptive Authorization through Transformer-Based Learning
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
Access control governs how entities interact with protected resources and remains a central pillar of modern cybersecurity. Classical authorization frameworks such as DAC, MAC, RBAC, and ABAC provide structured and interpretable policy foundations but depend heavily on manual policy and attribute engineering, making them increasingly difficult to maintain in large, dynamic environments. As organizational roles, contextual attributes, and system conditions evolve, policy drift, privilege accumulation, and configuration inconsistencies emerge, weakening security posture. Recent advances in machine learning (ML) offer the ability to automate authorization by learning predictive relationships from user, resource, and contextual metadata. However, progress is constrained by the scarcity of real-world datasets, inconsistent evaluation methodologies, and challenges associated with heterogeneous or incomplete attribute sets.Motivated by these limitations, this work focuses on a synthetic data generation and evaluation framework for access-control models. Leveraging domain-informed synthetic datasets that emulate realistic healthcare authorization conditions including role hierarchies, permission distributions, and anomaly patterns, we systematically evaluate transformer models against traditional machine-learning approaches. Results show that while tree-based ensembles perform strongly in low-noise, low-data environments, transformer-based and other neural architectures scale more robustly as data scale and complexity grows. These findings reinforce emerging evidence that modern attention-based models offer promising advantages for adaptive, data-driven authorization in real-world environments.
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 | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.003 |
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