Unsupervised Insider Threat Detection Using Multi-Head Self-Attention Mechanisms
Bibliographic record
Abstract
We propose an unsupervised insider threat detection system that learns normal user behaviors through audit data using neural networks equipped with multi-head self-attention mechanisms. The attention mechanisms learn precise normal user behaviors using large event windows. The key idea is to consider a sequence abnormal when it exhibits some events that are not likely to happen given the preceding events sequence. In addition, our method provides insights into detected threats through an event-based scoring system to facilitate threat understanding by security experts. Furthermore, the proposed solution does not require intensive audit logs pre-processing, such as manual domain knowledge-rich features extraction and data balancing. Thus, the proposed approach is easy to implement and use across organizations regardless of their expertise level. The proposed solution also provides the ability to detect threats in real-time. On the benchmark dataset CERT version 4.2, our solution, combined with hyperparameter search by Bayesian optimization, outperforms the previous state-of-the-art approaches with an area under the ROC curve of 97.23%, a recall of 91.51%, an accuracy of 93.13% and a false positive rate of 6.8%.
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.
How this classification was reachedexpand
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 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".