Infosec industry to drive machine learning spend surge says analyst

Amid the AI hype is a real chance to spot more anomalous behaviour, faster

The information security industry's rush to adopt machine learning will help businesses burn US$96 billion on big data, intelligence, and analytics by 2021, says research house ABI .

The report by lead number cruncher Dimitrios Pavlakis claims User and Entity Behavior Analytics (UEBA) and "deep learning algorithm designs" will be widely adopted by security companies as they collectively put big data to work detecting threats.

The former machine learning technology, UEBA, is correlation on steroids, capable of detecting anomalies that can indicate if staff logins have been compromised and are being tested across the enterprise network.

It can learn the activities and services most typical of a user to generate alerts when something anomalous occurs, like login attempts to odd network shares. Vendors are buying up across the space including Splunk's buy of Caspida, and Arksight selling Securonix.

Antivirus vendors, Pavlakis says, are contributing too. Cylance is pushing its fuzzy antimalware capabilities as a something seated in the much attributed but difficult to acquire artificial intelligence space, for example.

Additionally," … the cyber security industry is investing heavily in machine learning in hopes of providing a more dynamic deterrent," Pavlakis says.

“This will drive machine learning solutions to soon emerge as the new norm beyond Security Information and Event Management, and ultimately displace a large portion of traditional antivirus, heuristics, and signature-based systems within the next five years.”

Pavlakis says signature-based antivirus will be "absorbed completely" into machine learning technology and agrees with wider analyst predictions that SIEM logging will be cleaved off and woven into UEBA.

Vendors including Gurucul; Niara; Splunk; StatusToday; Trudera, and Vectra Networks are wannabe UEBA innovator leaders in a market that counts Deep Instinct and Spark Cognition as entrants bearing feature-agnostic models, deep learning, and natural language processing, he says. ®