This is supporting code for the blog post (link TBD) where we use malware classification scenario to demostrate the point how useful Hierarchical Attention Networks (HANs) are for sequences analysis.
Folder csdms contains code related to CSDMC 2010 API malware vs benign software API class classification.
Fodler kaggle contains materials for binary malware classification. We used corpus provided in this Kaggle competition: Microsoft Malware Classification Challenge (BIG 2015). Note: we've used data only for two classes (Class 1: Ramnit and Class 2: Lollipop).