Hundreds of thousands of computers in 150 countries have been hit by the WannaCry ransomware. While users are scampering around trying to fix their computers, the top of mind question is whether this could have been avoided. And if artificial intelligence could have predicted and prevented such an attack.
At GPU Technology Conference (GTC) in San Jose, California last week, Israeli firm Deep Instinct won the Most Disruptive Startup category in NVIDIA’s Inception Award. The firm is the first to use AI to predict and prevent malware attacks.
According David Eli, Chief Technology Officer (CTO) of Deep Instinct, more than a million new malware threats are released daily, but most antivirus software focuses on known threats.
His firm’s graphics processing unit (GPU)-accelerated deep learning software detects malware in real time. Trained on hundreds of millions of files, the neural network learns to detect more threats and then uses its experience to predict new attacks.
“Winning this prize is the ultimate recognition from the deep learning industry because deep learning and NVIDIA are synonymous,” said Eli.
This is the inaugural year of NVIDIA’s Inception Awards, which recognises startups in three categories – Hottest Emerging, Most Disruptive and Social Innovation. Winners received significant cash prizes and graphics processing unit (GPU) hardware to further accelerate their activities.
Emerging out of stealth mode in November 2015, Deep Instinct’s patent-pending application of deep learning to cybersecurity results in cutting-edge capabilities of unmatched accurate detection and real-time prevention.
Leveraging the capabilities associated with deep learning, Deep Instinct provides instinctive protection on any device, platform, and operating system. Zero-day and APT attacks are immediately detected and blocked before any harm can happen to the enterprise’s endpoints, servers, and mobile devices.
“Deep Instinct relies on end-to-end deep learning for all its advanced malware detection and prevention capabilities. The deep neural network is trained on hundreds of millions of malicious and legitimate files. To handle such large-scale training, Deep Instinct developed its proprietary deep learning infrastructure directly on NVIDIA’s GPU machines,” said Eli.
“The powerful capabilities of NVIDIA GPUs enable us to perform our training at a substantially faster speed compared to CPUs: while training the Deep Instinct brain on NVIDIA’s GPUs takes a little over a single day of training, the same task on CPUs would take more than three months,” he added.
“We are thrilled to be recognised by NVIDIA for what we believe is a groundbreaking application of GPUs. Being able to leverage powerful technological capabilities to apply deep learning to cybersecurity empowers enterprises with unprecedented, real-time protection from the next unexpected attack,” said Guy Caspi, Chief Executive Officer of Deep Instinct.