Security analytics are only as good as the data that feeds them. I’m not saying that advanced analytics such as machine learning and artificial intelligence (AI) aren’t amazingly powerful and increasing important for cybersecurity — they are. But unlocking the full potential of advanced analytics requires deep network insights: a combination of the right data, sufficient visibility into the details and context that we care about from a security perspective.
The Power of Deep Learning
I was reminded of this recently when I came across a research article on the use of deep learning for speech recognition. We’ve all witnessed firsthand just how good speech recognition has become, and it’s fundamentally changing how we interact with devices. Just look at the smart speakers finding their way into our homes. All we have to do is deploy these small cylindrical devices around the house, connect them and they will respond to voice commands, enabling us to play our favorite music, change the temperature on our smart thermostat or even order those last-minute holiday presents for family members.
However, this technology is dependent on being able to pick up our voices with sufficient audio clarity to discern what is being said. For this reason, these devices are designed to have multiple sensors deployed at different locations around our homes to detect what is spoken. After all, these devices are useless if they miss the phrase that begins a command.
Diving Deep Into Data
It’s no different in the security space. Analytics give us the power to find anomalous activity that humans simply couldn’t do in any reasonable amount of time. But this assumes that we have the data with the detail and clarity that we need in the first place.
We’ve been focusing heavily on deep network insights and data extraction with QRadar Network Insights to extend the capabilities of our security analytics. This combination allows us to go way beyond trying to determine good versus bad the moment data crosses the network. Instead, we can leverage a comprehensive set of analytics to look at detected content and extracted data in the context of their applications, across many disparate network sessions and over time to identify suspect activity that would otherwise go unnoticed.
A great example of this is user behavior analytics (UBA). UBA goes far beyond traditional rules by looking at behavior over time and adjusting risk scores based on activity associated with individual users. So it’s not surprising that the more granular we can become in term of visibility and insights into the activity of a given user, the more specific we can be in the risk scores we assign. Now, uploading sensitive company data to unsecured file share applications looks very different from sharing inconsequential information on social media — and this is only the start.
The Benefit of Deep Network Insights
The network inherently contains the data we need to generate the insights and security context required for next-generation analytics. The key is to unlock them with UBA and other processes designed to identify patterns that could point to potentially malicious behavior. This makes our smart devices smarter and our security tools more effective at monitoring the nuances of user activity to detect threats before they spin out of control.
Watch the on-demand webinar: Using Network Visibility to Detect Well-Hidden Threats
QRadar Product Manager, IBM
Tom Obremski is a product manager for IBM QRadar Security Intelligence. Tom has over 20 years of experience developing and bringing technology products to ma...