Interactive Machine Learning for Data Exfiltration Detection: Active Learning with Human Expertise
Bibliographic record
Abstract
Data exfiltration is a serious threat to organizations. Such exfiltrations cause breach events that can lead to millions of dollars of loss. Perimeter defense is not enough by itself since successful exploits from insiders can also be very damaging. Internal network user activities need to be monitored to detect malicious actions. Automatic machine learning methods can be applied for network anomaly detection, but they create a lot of false alarms. Domain experts can identify malicious users, but they are unable to process large volumes of data. Interactive machine learning (iML) deals with this tradeoff by creating an efficient collaboration between domain experts and machine learning algorithms. Previous research in iML has focused mainly on collaboration with non-experts. The design and requirements for expertise-driven iML have yet to be delineated for cybersecurity applications. In this research, we proposed an Active Learning (AL) model trained with outputs from a liberal (outputting many false alarms as well as possible hits) anomaly detection (AD) criterion to study expert-iML collaboration in anomaly detection. The results showed that: iML in this context can prune false alarms and minimize misses; the performance/compatibility tradeoff that typically occurs in conventional machine learning updates may be less salient in iML. We suggest that compatibility between experts and algorithms can be improved by presenting information about feature relevance during the training process.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.007 | 0.019 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.002 | 0.004 |
| Open science | 0.003 | 0.003 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.002 | 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 source (direct Gemma or distilled Codex), 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".