Malware is one of the major problems that are plaguing the digital world. In the “best” case it will only affect one’s private computer, in the worse case it may affect a whole business network and easily cause losses in the six figures range. To recognize malware swiftly is therefore one of the major features that need to be in place with every security software out there.
Microsoft and Intel got together in order to find a better way how to identify malware. The project name Stamina stands for a rand new technology that’s able to turn malware into pictures. Pictures? What exactly does this mean? As much as it may be exciting mostly for nerds only, the reason is not an artistic one.
Artificial intelligence is going through the generated pictures and searches for patterns that hint on malware. Deep learning is essential in the process. Stamina stands for “Static Malware -as-image Network Analysis”, transforms the regular and malicious code alike into pictures. From raw pixel data is created a two-dimensional picture, and an algorithm then searches for patterns.
The size of each picture depends on the amount of data coming in, each is however reduced into what may be quickly searched through for suspicious patterns. There has been 2,2 million unique codes analyzed as a part of the “Stamina” project, and the malware has been eventually recognized with a a stellar success ratio of 99,07 % cases! Something that sounds very promising in terms of commercial use.