News
News Categories

Your future anti-virus software may not require any updates at all

By Koh Wanzi - on 2 Nov 2015, 12:21pm

Your future anti-virus software may not require any updates at all

Image Source: Pixabay

Deep learning may be all about helping computers become smarter – by mimicking biological neural networks – but we are only beginning to discover the myriad possibilities that it opens up.

As it turns out, security researchers have found that deep learning algorithms can help anti-virus software identify new malicious code, even without database updates.

Computers are traditionally bad at things like pattern recognition, but deep learning algorithms and artificial neural networks can help them become better. And since viruses, trojans and other malicious code often share similarities among each other, it makes sense that anti-virus software could benefit from deep learning, which would help them recognize certain traits in brand new threats.

An Israeli startup called Deep Instinct is reportedly building an anti-virus suite that can flag new malware with 20% better accuracy than the best anti-virus suites today. Although Deep Instinct’s claims have yet to be verified, there’s reason to believe that they are credible. Microsoft and Invincea have both published research papers that demonstrate the effectiveness of applying deep learning algorithms to malware detection, with one paper reporting a 95% success rate at detecting malware without updates.

In addition, back in September, Qualcomm announced a new malware behaviour analysis featured called Qualcomm Snapdragon Smart Protect, which would leverage its upcoming Snapdragon 820 processor and Zeroth neural chip to provide better protection against malware for mobile devices.

The threat landscape is rapidly evolving, and deep learning could just help security solutions stay ahead of the curve and even minimize exploitation of zero-day vulnerabilities. 

Source: MIT Technology Review via Engadget

Join HWZ's Telegram channel here and catch all the latest tech news!
Our articles may contain affiliate links. If you buy through these links, we may earn a small commission.