THREATIFY: APT Threat Variant Generation Using Graph-Based Machine Learning
Bibliographic record
Abstract
Ensuring cybersecurity in an ever-evolving threat landscape requires proactive identification and understanding of potential threats. Conventional detection and prediction solutions often fall short as they predominantly focus on known attack vectors. Advanced Persistent Threats (APTs) are becoming increasingly sophisticated and stealthy, resulting in new threat variants that are undetectable by these detection solutions. This paper introduces THREATIFY, a novel approach to predicting the most probable threat variants from existing APTs and previously seen attack campaigns. Our approach automates the generation of threat variants using graph-based machine learning based on the attack definition, past attack campaigns, and the security context between different techniques. THREATIFY leverages a security knowledge base of realistic attack scenarios and cybersecurity expertise to model, generate, and predict new forms of potential future threats by combining inter-(i.e. within the same APT attack) and intra-(i.e. between different APTs) techniques used by threat actors. It is crucial to emphasize that THREATIFY does not merely mix techniques from different APTs; rather, it constructs a logical and pragmatic kill chain based on their security context. THREATIFY is able to predict new attack steps, find relevant techniques to be substituted by, and merge APTs techniques in the current security context, and thus create previously unexplored threat variants. Our extensive experimental results demonstrate the efficacy of our approach in generating relevant and novel threat variants with a similarity score of 92%, uniqueness of 82%, validity of 95%, and reduction rate of 96%, including those that have never occurred before.
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 | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| 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".