System and method for analyzing binary code for malware classification using artificial neural network techniques
A system for detecting whether a file including content s associated with a cyber-attack is described. The content may include an executable file for example. The system includes an intelligence-driven analysis subsystem and a computation analysis subsystem. The intelligence-driven analysis subsystem is configured to (i) receive the file, (ii) inspect and compute features of the file for indicators associated with a cyber-attack, and (iii) produce a first output representing the detected indicators. The computational analysis subsystem includes an artificial neural network to (i) receive a network input being a first representation of at least one section of binary code from the file as input, and (ii) process the first representation of the section to produce a second output. The first output and the second output are used in determination a classification assigned to the file.
1. A cyber-security system configured to perform malware classification using neural networks, the cyber-security system comprising:
processing circuitry and
one or more non-transitory computer-readable storage media that store:
a convolutional neural network; and
instructions for performing operations, the operations comprising:
obtaining binary code of an executable file;
processing at least a portion of the binary code of the executable file with the convolutional neural network to generate, as an output of the convolutional neural network, a network output; and
processing the network output with a classifier to generate a threat score for the executable file, wherein the threat score indicates whether the executable file comprises malware, wherein the threat score is generated by threat assessment logic of the classifier. wherein the threat assessment logic performs a sigmoid function to normalize the threat score as a scalar value within a prescribed value range.
2. The cyber-security system of claim 1 , wherein the convolutional neural network has been previously trained using labeled training sets of malicious and benign binary code files.
3. The cyber-security system of claim 1 , wherein the convolutional neural network comprises a plurality of executable logic layers comprising one or more convolution layers, one or more pooling layers, and one or more fully connected/nonlinearity (FCN) layers.
4. The cyber-security system of claim 1 , further comprising a pre-processor configured to select the portion of binary code from the executable file for analysis by the convolutional neural network.
5. The cyber-security system of claim 4 , wherein the pre-processor is configured to select a plurality of subsections of the binary code for analysis by the convolutional neural network, each subsection being combined and conditioned for analysis.
6. The cyber-security system of claim 1 , wherein the executable file is classified as malicious when the scalar value exceeds a threshold value within a prescribed range.
7. One or more non-transitory computer-readable storage media that store:
a convolutional neural network; and
instructions for performing operations, the operations comprising:
obtaining binary code of an executable file;
processing at least a portion of the binary code of the executable file with the convolutional neural network to generate, as an output of the convolutional neural network, a network output; and
processing the network output with a classifier to generate a threat score for the executable file, wherein the threat score indicates whether the executable file comprises malware, wherein the threat score indicates whether the executable file comprises malware, wherein the threat score is generated by threat assessment logic of the classifier, wherein the threat assessment logic performs a sigmoid function to normalize the threat score as a scalar value within a prescribed value range.
8. The one or more non-transitory computer-readable storage media of claim 7 , wherein the convolutional neural network has been previously trained using labeled training sets of malicious and benign binary code files.
9. The one or more non-transitory computer-readable storage media of claim 7 , wherein the convolutional neural network comprises a plurality of executable logic layers comprising one or more convolution layers, one or more pooling layers, and one or more fully connected/nonlinearity (FCN) layers.
10. The one or more non-transitory computer-readable storage media of claim 7 , further comprising a pre-processor configured to select the portion of binary code from the executable file for analysis by the convolutional neural network.
11. The one or more non-transitory computer-readable storage media of claim 10 , wherein the pre-processor is configured to select a plurality of subsections of the binary code for analysis by the convolutional neural network, each subsection being combined and conditioned for analysis.
12. The one or more non-transitory computer-readable storage media of claim 7 , wherein the executable file is classified as malicious when the scalar value exceeds a threshold value within a prescribed range.
13. A computer-implemented method for malware classification using neural networks, the method comprising:
obtaining, by a computing system, binary code of an executable file;
processing, by the computing system, at least a portion of the binary code of the executable file with a convolutional neural network to generate, as an output of the convolutional neural network, a network output; and
processing, by the computing system, the network output with a classifier to generate a threat score for the executable file,
wherein the threat score indicates whether the executable file comprises malware, wherein the threat score is generated by threat assessment logic of the classifier, wherein the threat assessment logic performs a sigmoid function to normalize the threat score as a scalar value within a prescribed value range.
14. The computer-implemented method of claim 13 , wherein the convolutional neural network has been previously trained using labeled training sets of malicious and benign binary code files.
15. The computer-implemented method of claim 13 , wherein the convolutional neural network comprises a plurality of executable logic layers comprising one or more convolution layers, one or more pooling layers, and one or more fully connected/nonlinearity (FCN) layers.
16. The computer-implemented method of claim 13 , further comprising a pre-processor configured to select the portion of binary code from the executable file for analysis by the convolutional neural network.
17. The computer-implemented method of claim 16 , wherein the pre-processor is configured to select a plurality of subsections of the binary code for analysis by the convolutional neural network, each subsection being combined and conditioned for analysis.