Malware family classification via efficient Huffman features

As malware evolves and becomes more complex, researchers strive to develop detection classification schemes that abstract away from the internal intricacies of binary code represent without need for architectural knowledge or invasive analysis procedures. Such approaches can reduce complexities feature generation simplify process. In this paper, we present efficient Huffman features (eHf), a novel compression-based approach construction, based on encoding, where are represented in compact format, intrusive reverse-engineering dynamic processes. We demonstrate viability eHf as solution classifying into their respective families large corpus 15 k samples, indicative current threat landscape. evaluate against alternatives show our method is comparable superior accuracy, while exhibiting considerably greater runtime efficiency. Finally resilient reordering obfuscation.