| --- |
| license: cc-by-4.0 |
| task_categories: |
| - graph-ml |
| tags: |
| - android |
| - malware-detection |
| - security |
| - call-graphs |
| size_categories: |
| - 10K<n<100K |
| --- |
| |
| # AndroSecNet |
|
|
| Function call graphs extracted from **N** Android applications |
| (**N_mal** malware / **N_ben** benign) for graph-based malware detection. |
|
|
| ## Dataset Structure |
| - Format: WebDataset (`graphs.tar.gz`, ~60 GB) of gzipped pickled |
| Python `igraph.Graph` objects. |
| - Each sample is keyed by the app's SHA-256 hash, under |
| `graphs/malware/<sha256>` or `graphs/benign/<sha256>`. |
| - One `train` split; there is no predefined validation/test split. |
|
|
| ## Loading |
| ```python |
| import webdataset wds # (fix this line: `import webdataset as wds`) |
| # working loading snippet here |
