Organizations: Department of Electronics, Information and Bioengineering. Politecnico di Milano - Milan, Italy · Cleafy S.p.A. - Milan, Italy · Faculty of Informatics. Università della Svizzera Italiana - Lugano, Switzerland
Malware are malicious programs that are grouped into families based on their penetration technique, source code, and other characteristics. Classifying malware programs into their respective families is essential for building effective defenses against cyber threats. Machine learning models have a huge potential in malware detection on mobile devices, as malware families can be recognized by classifying permission data extracted from Android manifest files. Still, the malware classification task is challenging due to the high-dimensional nature of permission data and the limited availability of training samples. In particular, the steady emergence of new malware families makes it impossible to acquire a comprehensive training set covering all the malware classes. In this work, we present a malware classification system that, on top of classifying known malware, detects new ones. In particular, we combine an open-set recognition technique developed within the computer vision community, namely MaxLogit, with a tree-based Gradient Boosting classifier, which is particularly effective in classifying high-dimensional data. Our solution turns out to be very practical, as it can be seamlessly employed in a standard classification workflow, and efficient, as it adds minimal computational overhead. Experiments on public and proprietary datasets demonstrate the potential of our solution, which has been deployed in a business environment.
Figures & tables
Fig. 1 : Depiction of our proposed open-set classifier K .
Fig. 2 : Analogy between the Logits vector in a neural network-based classifier (a) and the decision values in a tree-based Gradient Boosting classifier (b). In both cases, the vector zp represents the raw output scores before applying the softmax function σ(zp) .
Fig. 3 : Imbalance in the cardinality # of each malware family ℓ in our proprietary dataset, where malware families with less than 10 samples are represented by pale-colored bars.
TABLE I : Micro-average (accuracy) and macro-average recall for closed-set C and open-set K classifiers on both public and proprietary datasets. The results are shown for the two tested settings Fixed-Novel (a) and LOO-Novel (b).
Fig. 4 : Recall confusion matrix of the open-set classifier K , on the public (first row) and proprietary (last row) datasets, when (a),(d) instances of the less populous classes are grouped in the dummy class others and treated as novel , and (b),(e) each class is sequentially treated as novel in the leave-one-class-out process, along with the associated novelty detection ROC curves (c),(f).