This repository contains details about MIL-based malware representation proposed in paper: "Stiborek, Jan and Pevný, Tomáš and Rehák, Martin: Multiple Instance Learning for Malware Classification, ESWA 2017 (revision submitted)".
Due to the privacy policy we are currently not able to release the complete dataset used in evaluation of the method proposed in the paper. However, we provide lists of SHA256 of samples along with their labels (samples_train.txt, samples_test.txt).
Due to the fact that the method described in the paper is part of the products shipped by CISCO Systems, Inc. we cannot currently release the complete source code of the proposed method.