Revisiting Temporal Inconsistency and Feature Extraction for Android Malware Detection
Bibliographic record
Abstract
The growing popularity and ease of access have turned Android applications into prime targets for malicious attackers. Within the security research community, machine learning has become an essential instrument for conducting Android malware detection and analysis. However, there are potential threats to validity of existing studies, mainly resulting from their used datasets. One of the primary issues is temporal inconsistency (also called temporal bias) that is caused by incorrect time splits of training and testing sets or using imprecise indicators for release time of apps. This paper investigates the use of Google Play Store upload year of an app as a precise indicator of its release time to address temporal bias in machine learning-based Android malware detection. Using this approach is made possible by AndroZoo’s December 2023 data release. Through a three-layer filtering process, we demonstrate the unreliability of the commonly used dex_date as the release time of an app and propose a more accurate approach for creating temporally-consistent datasets based on an app’s upload year. Additionally, we have open-sourced our data and feature extraction process for Android malware analysis, supporting both server-side and on-device extraction, to enhance research reproducibility and facilitate community access.
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 | 0.009 | 0.049 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.006 | 0.005 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.004 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
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".