Deep learning — taking the battle to malware

Most anti-virus software depends upon updates that allow it to recognize any possible threats. Unfortunately, it’s been like that for as long as we’ve had computers, and it’s a system that has never seen any real improvement. And with each day that passes, malware is becoming more sophisticated at figuring out how to bypass these updates to infect our computer systems.

Alas, there is finally a new technology being developed called “deep learning”, and it is promising unrivaled protection for our computers. 

One of those pushing the technology is an Israeli company called Deep Instinct, which is looking to change how anti-virus software protects our computers. 

The idea is that artificial neural networks are trained to recognize the characteristics of code looking to do harm. The neural networks are essentially large swaths of simulated neurons and synapses that are capable of recognizing patterns. These deep learning systems can be trained by using thousands of photos and then recognize that person in newer photos. It’s safe to say that this software is the digital version of a bloodhound.    

According to technologyreview.com, the best malware detection software is only good for detecting upwards 87% of new threats. Another system developed by two Invincea employees can detect and stop upwards of 95% of threats with a 1% error rate. 

This new way to teach our computers a better way to learn is also being used for other parts of the security sector. Areas like handwriting recognition and voice recognition are but a couple of ways this technology is transforming digital security software. As it grows, so too will the companies who use it. Invincea already has clients such as Dell and MGM Resorts, where will we see deep learning used next? 

Source: MIT Technology Review

By Aaron LaBarbera