The Deep Learning Revolution: A Cybersecurity Use Case

Eli David, Co-Founder & CTO, Deep Instinct

The year 2017 was a turning point in global cybercrime, in both volume of attacks and the cost of damages. According to a report by Wells Fargo this year, cybersecurity is the number-one issue keeping CEOs up at night, but traditional reactive approaches are not working. Although the vast majority of new cyberthreats are very small mutations of known threats, current cybersecurity approaches are not able to recognize these small changes.

The kind of feature recognition that is simple for humans—such as telling the difference between a dog and a cat in a photograph—presents huge technological challenges for machine learning. That’s because machine learning depends on extracting features from an image or dataset and then learning rules about those features.

The solution is to turn to deep learning, a subfield of machine learning. Deep learning skips the feature extraction step entirely, instead feeding pixel values directly into a deep neural network. Results obtained by deep learning represent the biggest leaps in performance in the history of AI.

The benefits of deep learning for cybersecurity include real-time detection and prevention, real-time classification, prediction of unknown (future) threats, operation on any device or operating systems, and connectionless edge deployment. Future development of deep learning technologies will continue to advance its usefulness for cybersecurity.

Be the first.

At Samsung, we are passionate about fostering innovation and we'd love to keep in touch with you about the latest updates in artificial intelligence, mobility, digital health, and more. You may unsubscribe at any time.  For more information on our commitment to privacy, please see our privacy policy.