Online fraudsters in the cross hairs of Featurespace sharpshooters
Life has just got tougher for online financial fraudsters after Cambridge-based Featurespace added two powerful weapons to its online defence armoury.
The company has filed two gilt-edged global patents that will enable new levels of customer protection across the financial industry.
The first is for Featurespace’s Automated Deep Behavioural Networks for the card and payments industry.
Automated Deep Behavioural Networks are a deep neural network architecture of connections and updates that recognise and prevent significantly more fraud cases.
Deep neural networks have revolutionised areas such as image recognition and text understanding by creating specific architectures (connections and weights) designed to extract meaning from the underlying data presented to the network.
Automated Deep Behavioural Networks solves the problem of finding a neural net architecture that extracts meaning from transaction sequences producing a much higher distinction between genuine and fraudulent transactions.
Featurespace co-founder Dave Excell said: “Financial institutions around the world are experiencing fraud and account takeover at unprecedented levels, with some reports estimating this number at more than $40 billion last year.
“The Automated Deep Behavioural Networks patent and associated technologies deliver the right levels of model performance the industry needs to decrease fraud and protect consumer accounts before attacks happen.”
Featurespace’s second patent is for Behavioural Anomaly Score, which identifies anomalies in individual customer behaviour without having any prior knowledge of contextual high-risk behaviour.
This technology appreciably amplifies the ability to identify when a person’s behaviour is out of character without any labelled data.
Through a Behavioural Anomaly Score, companies and financial institutions can see the exact point at which a person’s behaviour has changed with greater precision and from there, construct more complex models for change detection – further reducing the incidences of financial crime.
Excell said: “The fight against fraudsters and the organisations that commit financial crime on a large scale is challenging and ever-evolving.
“Technology, specifically machine learning, will continue to be central in this fight and these two patents from Featurespace advance our leading market position and our capacity to help progressive financial institutions protect the consumer.”
Identity fraud cost US citizens alone a total of around $56 billion last year, with some 49 million consumers falling victim, according to a study by Javelin Strategy & Research.
This side of the Pond, a UK Citizens Advice report states one in three adults in the UK were targeted by a pandemic-related scam last year.
Featurespace says full public patent applications have been filed in the US, UK, EU and within the Patent Cooperation Treaty – a unified procedure for filing patent applications to protect inventions in each of PCT’s contracting states.