2017 · 108 citations · 14 references
ComputingComputer ScienceCybersecurityComputer SecurityMachine LearningMalware AnalysisDeep LearningIntrusion DetectionMalware ClassificationMalware IntelligenceRansomwareThreat DetectionDeep Learning For Malware ClassificationDeep Learning-based Malware DetectionMachine Learning For Malware ClassificationMachine Learning-based Malware DetectionMultilayer PerceptronPerceptronRansomware ClassificationRansomware Detection
Ransomware covertly installs and executes cryptovirology attacks, and its prevalence has surged recently, inflicting tens of millions of dollars in losses. The study evaluates shallow and deep neural networks for detecting and classifying ransomware. The authors use API invocation patterns to distinguish ransomware from benign and other families, training multi‑layer perceptrons with varied architectures over 500 epochs and learning rates between 0.01 and 0.5. The MLP achieved perfect accuracy (1.0) distinguishing benign .EXE files from ransomware and 0.98 accuracy in classifying ransomware families, outperforming classical machine‑learning classifiers.
Ransomware is one type of malware that covertly installs and executes a cryptovirology attack on a victims computer to demand a ransom payment for restoration of the infected resources. This kind of malware has been growing largely in recent days and causes tens of millions of dollars losses to consumers. In this paper, we evaluate shallow and deep networks for the detection and classification of ransomware. To characterize and distinguish ransomware over benign and various other families of ransomwares, we leverage the dominance of application programming interface (API) invocations. To select a best architecture for the multi-layer perceptron (MLP), we done various experiments related to network parameters and structures. All the experiments are run up to 500 epochs with a learning rate in the range [0.01-0.5]. Result obtained on our data set is more promising to distinguish ransomware not only from benign from its families too. On distinguishing the .EXE as either benign or ransomware, MLP has attained highest accuracy 1.0 and classifying the ransomware to their categories obtained highest accuracy 0.98. Moreover, MLP has performed well in detecting and classifying ransomwares in comparison to the other classical machine learning classifiers.
14
Scikit-learn: Machine Learning in Python
Fabián Pedregosa, Gaël Varoquaux, Alexandre Gramfort et al. · arXiv (Cornell University) · 2012 · 63.7K citations · Full text
Corinna Cortes, Vladimir Vapnik · Machine Learning · 1995 · 40.5K citations · Full text
Laurens van der Maaten, Geoffrey E. Hinton · Journal of Machine Learning Research · 2008 · 35.7K citations
Corinna Cortes, Vladimir Vapnik · Machine Learning · 1995 · 33K citations · Full text
TensorFlow: A system for large-scale machine learning
Martı́n Abadi, Paul Barham, Jianmin Chen et al. · arXiv (Cornell University) · 2016 · 8.8K citations · Full text