X
Tech

SparkCognition: Let machines address security threats

Can machine learning, predictive analytics and big data analysis ferret out security threats before they can harm an organization's IT assets?
Written by Dan Kusnetzky, Contributor

A long-time industry contact, Amir Husain, reached out to me recently to let me know that he had moved on and was in the process of starting up a new company, SparkCognition. Husain and I have been chatting about advances in technology throughout his time as CTO at ClearCube and later as CEO of VDIworks.

This time, Husain has turned his attention to using a combination of cloud computing, big data, machine learning and predictive analytics to predict, find and resolve security threats. The idea has merit and I expect we'll see big things come of it over time.

Predictive analytics and machine learning

I've spoken with a number of companies that are using the huge volume of operational or machine log data that most organizations have been collecting —combined with the clever use of some very complex statistical models and machine learning techniques — to help address various types of management problems.

Typical targets of this technology have been performance management, end user experience management, predicting customer behavior or preventing credit card abuse.

Many of these suppliers offer their technology to customers through the use of a cloud-based service. This allows them to rapidly collect and analyze anonymous data to develop a model of a customer's current state. Then it is possible to quickly detect changes from that modeled state, determine the causes and recommend or implement changes to stem a potential problem.

SparkCognition's MindSpark

SparkCognition is planning to take this combination of statistical modeling, machine learning, big data analysis and cloud computing in a slightly different direction — finding and preventing security issues before they have a chance to become problems. SparkCognition's product is called MindSpark.

According to Husain, the MindSpark platform is built on patent-pending Pattern Recognition and Machine Learning techniques that enable  cognitive capability. He pointed out that MindSpark — when exposed to security data —  finds patterns of attack, identifies vectors, models attacker behavior, and much more.

Husain also said that MindSpark aggregates its learning at a faster pace than any human or legacy software system. What it learns — the statistics models and base operational data — is offered as a cloud service.

Husain expects that the service is very useful today and will only increase in capability as it gains more experience. MindSpark, he said, is likely to grow in both intelligence and capability over time and will shortly be able to address security questions that only a handful of elite security experts can address today.

Only time will tell if this approach will live up to Husain's vision. If his previous work can be used as a guide, SparkCognition has a very strong potential.

Editorial standards