Thought Leaders in Cyber Security: Mark Jaffe, CEO of Prelert (Part 5)

Sramana Mitra: If I understand correctly, you have this behavioral detection data anlaysis going on. The machine learning is correcting things. As new use cases pop up, the system administrators can set up new heuristics on what the machine learning algorithm should be doing in an unsupervised mode to correct those.

Mark Jaffe: It’s not to correct them as much as they can use our single platform to deploy new use cases within a minute. They can initially use the product to identify DNS tunneling accurately in their environment.

Sramana Mitra: They can identify a set of use cases that are causing these alerts. Then they come up with remedies for those. They can program those remedy elements in an unsupervised learning mode into your machine learning algorithm. Is that what you’re saying?

Mark Jaffe: Not really. What’s unsupervised is it can be against any use case. I used DNS tunneling as an example because it’s something you identify from DNS logs. Then, you want to look at your Firewall logs and your next use case is to look for data exfiltration. Next, you want to look at endpoints and understand unusual behaviors of endpoints. You want to look at server performance metrics and identify why machines are behaving slowly.

All of that is doable and is quickly enabled in our product because of the fact that, under the hood, we have unsupervised machine learning. All those other companies that are starting up are using supervised learning techniques, which means they have to write custom code to implement each one of those use cases. The time to deployment or time to modification is much higher.

This segment is part 5 in the series : Thought Leaders in Cyber Security: Mark Jaffe, CEO of Prelert
1 2 3 4 5 6