Over the years, malware has become one of the biggest threats to mankind majorly because of how it exists in the computer world and not many can spot it. However, Microsoft and Intel have just given a hope with their latest innovation that might make the situation better.
Both of the companies are now working on visualizing the malware through their collaboration with STAMINA (Static Malware-as-Image Network Analysis) - a project that is based on converting code into greyscale images just to make sure that deep learning systems could later study them.
Going deep into the further details of the approach, Microsoft and Intel are trying to make the binary form of an input file appear in a more simple stream of pixels which later would be turned into a picture with varying dimensions and file size. There is a trained neural network (currently under the testing phase) that will identify if anything affects the file.
Microsoft can collect large amounts of data from Windows Defender installations to train the AI system. The technology doesn’t need to fully recreate the virus with details in pixel because in case of large malware, translating the code into massive pictures doesn’t make sense.
STAMINA’s accuracy has been pretty great so far with 99% successful classification of malware and only 2.6% false-positive rates. But then at the same time, their system also becomes limited when it comes to dealing with large malware.
If the shortcomings can be solved refinement by Microsoft and Intel’s engineers, then we are definitely in for a big change. Majority of the malware depends heavily on the binary signatures or fingerprints. Hence, if the requirement turns up to getting sheer number of signatures, then trapping a user might just become very difficult.
With its effectiveness, anti-malware tools can also level up and minimize the chances of malware beating the defense systems provided by them.
Read next: Pirated Movies Might Give You Malware According to Microsoft
Both of the companies are now working on visualizing the malware through their collaboration with STAMINA (Static Malware-as-Image Network Analysis) - a project that is based on converting code into greyscale images just to make sure that deep learning systems could later study them.
Going deep into the further details of the approach, Microsoft and Intel are trying to make the binary form of an input file appear in a more simple stream of pixels which later would be turned into a picture with varying dimensions and file size. There is a trained neural network (currently under the testing phase) that will identify if anything affects the file.
Microsoft can collect large amounts of data from Windows Defender installations to train the AI system. The technology doesn’t need to fully recreate the virus with details in pixel because in case of large malware, translating the code into massive pictures doesn’t make sense.
STAMINA’s accuracy has been pretty great so far with 99% successful classification of malware and only 2.6% false-positive rates. But then at the same time, their system also becomes limited when it comes to dealing with large malware.
- Also read: How Gaming Cheats Lead To Malware
If the shortcomings can be solved refinement by Microsoft and Intel’s engineers, then we are definitely in for a big change. Majority of the malware depends heavily on the binary signatures or fingerprints. Hence, if the requirement turns up to getting sheer number of signatures, then trapping a user might just become very difficult.
With its effectiveness, anti-malware tools can also level up and minimize the chances of malware beating the defense systems provided by them.
Read next: Pirated Movies Might Give You Malware According to Microsoft
