Dark Reading is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them.Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.


11:55 AM
Rutrell Yasin
Rutrell Yasin
Connect Directly

5 Things Security Pros Need To Know About Machine Learning

Experts share best practices for data integrity, pattern recognition and computing power to help enterprises get the most out of machine learning-based technology for cybersecurity.
5 of 6

#4 Supervised or Unsupervised Learning? It Depends

There are two camps in the machine learning space: those who adhere to supervised learning and those that rely on unsupervised learning. Choosing the one best suited for your organization will depend on your resources and environment, experts say.

With Supervised Learning, analysts help train the system. Unsupervised Learning is autonomous; it uses a set of algorithms and learns from its dataset.

Just as every immune system is different, every network is different, says Justin Fier, director of cyber intelligence with Darktrace, a developer of self-learning software inspired by the biological principles of the immune system, which uses unsupervised machine learning.

Any adversary with the right amount of resources and patience can get through your perimeter. We take the approach -just as your immune systems does - of getting a sense of self, Fier says. Once deployed on a network, the companys Enterprise Immune System constantly learns what is normal for the network. From that, analysts can pull out the most minute anomaly the needle in the haystack of patterns that just dont belong, Fier says.

We do it in an unsupervised way, meaning we are not training the device. As far as learning what the devices are, the pattern of life, the different characteristics of the data we are ingesting for the modeling - all of that is done unsupervised without any human hand at training it.

I wouldnt say one [approach] is better than another, says Fier. It comes down to resources. Once, Fier deployed the technology for a proof-of-value at a company hosting a sporting event. The network administrator was scrabbling. He said he had laid enough fiber cable to go to the moon and back five times, Fier says. He didnt have the time to do a proof-of-value evaluation.

We plugged [the Darktrace tool] in and pointed data to the device and that was it. We didnt have to spend time building up configuration files or telling it what to do. It was already built-in and learns from data, Fier says.

Is one approach better than the other? It depends on the environment you want to deploy in. I would err on the side of doing unsupervised because I dont have to assign a team of people to set the things up and train it on the datasets, he says.

Image Source: By a-image via Shutterestock

5 of 6
Comment  | 
Print  | 
Oldest First  |  Newest First  |  Threaded View
User Rank: Apprentice
12/14/2016 | 10:11:16 AM
Machine Learning - Useful points
First of all, a big thanks for the article. The informations (5 security pros) mentioned in this article very useful. As AI and ML is going to rule future world, we need to consider these security pros.
User Rank: Author
12/15/2016 | 3:02:27 PM
Insightful, thank you for sharing.
COVID-19: Latest Security News & Commentary
Dark Reading Staff 11/19/2020
How to Identify Cobalt Strike on Your Network
Zohar Buber, Security Analyst,  11/18/2020
New Proposed DNS Security Features Released
Kelly Jackson Higgins, Executive Editor at Dark Reading,  11/19/2020
Register for Dark Reading Newsletters
White Papers
Cartoon Contest
Write a Caption, Win an Amazon Gift Card! Click Here
Latest Comment: A GONG is as good as a cyber attack.
Current Issue
2021 Top Enterprise IT Trends
We've identified the key trends that are poised to impact the IT landscape in 2021. Find out why they're important and how they will affect you today!
Flash Poll
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
PUBLISHED: 2020-11-23
A user authorized to perform database queries may trigger denial of service by issuing specially crafted queries, which loop indefinitely in mathematics processing while retaining locks. This issue affects: MongoDB Inc. MongoDB Server v4.0 versions prior to 4.0.5; v3.6 versions prior to 3.6.10; v3.4...
PUBLISHED: 2020-11-23
Use after free vulnerability in EDK II may allow an authenticated user to potentially enable escalation of privilege, information disclosure and/or denial of service via adjacent access.
PUBLISHED: 2020-11-23
Logic issue EDK II may allow an unauthenticated user to potentially enable denial of service via adjacent access.
PUBLISHED: 2020-11-23
Out of bounds write in Intel(R) PROSet/Wireless WiFi products on Windows 10 may allow an authenticated user to potentially enable denial of service via local access.
PUBLISHED: 2020-11-23
Improper input validation in BlueZ may allow an unauthenticated user to potentially enable escalation of privilege via adjacent access.