May 12, 2020, 1:28 am
Note: We used commission links in this article and marked them with “*”. If an order is placed via these links, t3n.de receives a commission.
Microsoft and Intel are currently working to better analyze malware. The project uses a new technology that converts malware into images.
Tech giants Microsoft and Intel are working together to fight malware. As part of the project called Stamina (Static Malware-as-Image Network Analysis), malicious code is converted into black and white images. An artificial intelligence (AI) is to use deep learning to search the images for specific structural patterns that are specific to the respective malware example, such as ZD-Net reports.
Malicious code becomes a 2D image
The results of the corresponding research have the Intel and Microsoft researchers in a white paper published to Stamina. The conversion of the malicious code into an image takes place in several steps. First, the binary data is converted into raw pixel data. A two-dimensional image is then generated from the one-dimensional pixel stream. This can then be examined using conventional image analysis algorithms.
Stamina – new technology from Microsoft and Intel converts malware into images. (Photo: Intel)
–
The width of the image depends on the size of the original file. The height in turn is determined depending on the width and the raw data. The researchers reduced the size of the image because otherwise the images, which could quickly reach several billion pixels in size, could not be processed quickly enough. The researchers were convinced that the reduction should not have a negative impact on the result.
Stamina for malware classification
Microsoft and Intel trained the deep learning system with 2.2 million infected files in advance of corresponding tests. Stamina was then able to detect malware files with an accuracy of 99.07 percent. The researchers believe that the system can now be used for malware classification.
More on the subject:
