What these two words really mean and how they can save your company time and money – you just need to ask the right questions.
The concept of machine learning is everywhere in the cybersecurity industry. We might not yet have machines that can mimic human thinking in the way popular culture has been predicting for decades, but we have something that is getting close; technology that can learn. In the same way students improve the more information they receive about a subject, machine learning improves in real time with every piece of data it receives.
Machine learning has the potential to revolutionize a plethora of industries from advertising to retail, but the stakes are higher for security teams. The global shortage of cybersecurity talent to keep up with the ever-widening attack vector shows no signs of slowing down – in fact, it’s likely to get significantly worse in the coming years, with estimates suggesting that by 2019 there could be 1.5 million unfilled cyber jobs worldwide.
But what is machine learning? This concept has been added to many product descriptions, often to make them sound cool, cutting-edge or just ‘smart.’ Machine learning is when a machine can progressively improve performance on a specific task, without being explicitly programmed.
Now, the two questions you need to ask when something has machine learning capabilities are: What type of data is feeding this machine? And, who is teaching the machine?
At the end of the day, machine learning is an application of AI. As Andrew Ng, head of Baidu AI Group and Google Brain puts it, “AI is akin to building a rocket ship. You need a huge engine and a lot of fuel. The rocket engine is the learning algorithms but the fuel is the huge amounts of data we can feed to these algorithms.”
The intelligence behind machine learning – the algorithms and the data – is just as important as the output of the machine. These are the key elements you need to know to find out how powerful an application of machine learning really is.
As an example, our brain is a self-learning machine and yet, if we are sent to a tribe deep in the Amazonas, we won’t understand what they say unless someone gives us some words and grammar structures (data and algorithms) so we can, in time, put them together and follow conversations. Machine learning needs exactly the same, a good chunk of data and algorithms to understand the new information coming at it and learn. Without that those, machine learning is just an empty a buzzword. The applications of machine learning are especially relevant in the cybersecurity space.
When used to its full potential, machine learning can drastically reduce many current cybersecurity threats while cutting down on operational costs. It can be programmed to work through endless user signals, learn the difference between anomalous and malicious activity, and cut down manual reviews. What’s more, machine learning can improve its ability to aid security teams by turning what would once have been a potential disaster into a learning opportunity.
For example, once a fraudulent transaction is detected, the system can gather that transaction’s signs and analyze them. Then, the next time a similar fraudulent transaction is attempted, the system will flag it right away before any fraud happens. Even though machine learning is not a silver bullet against fraud, it’s helping companies thwart attacks that would have previously taken longer to detect. The power of machine learning lies on the intelligence behind it: this is the real information you need to find out before being lured by those two catchy words.
Related to this post: Fraud predictions for 2018 – Start planning your new year security resolutions
Authenticating on today’s breached world? Watch our webinar featuring Forrester analyst firm.
Want to read more posts like this? See our full blog here.