Kate O'flaherty | sc magazine | 21 april 2016
PA’s Nick Kotsis, a business intelligence expert, is quoted in an article on big data and the use of machine learning to spot anomalous behaviour.
The article explains that ‘machine learning' techniques are being developed as data that is being captured by businesses is becoming far too expansive for humans to analyse for unusual activity. Big data systems can pick up on attacks in real time and analyse data into something meaningful that can be interpreted by businesses.
The article goes on to explain that a machine learning system has the ability to alert a human to take action. Nick says: “This happens via pattern cognition allowing the system to discriminate between a typical action and an abnormality.”
He goes on to say that an unknown user, for example, might try to access a company system. A machine learning system will pick this out and raise it as an exception to the norm because it has been “trained” to separate suspicious behaviour from normal activity.
Nick concludes: “So, if you make a prediction on what might happen if a series of events come together, when the machine assumes something is wrong, it will notify someone.”