Our Technology

Nt's technology is the culmination of almost 20 years' extensive research and development into artificial intelligence and advanced neural processes. Using this comprehensive knowledge and experience, the company has developed a powerful, proprietary neural architecture that sets it apart from the competition.

Nt employs this unique neural architecture in combination with a variety of techniques to offer a thorough understanding of both customer and transactional activity to enable corporations to make informed risk management decisions. Accuracy is further enhanced by individual solution configuration to ensure each organisation's specific business requirements are met.

Neural Networks

Neural networks are ideal for handling large amounts of data and recognising patterns within complicated, vague or incomplete information. A neural network stores information in much the same ways as the human brain, i.e. each piece of data is inter-related to each and every other piece. This facilitates the identification of meaningful and predictive trends and patterns within data, and highlights anomalies that may be indicative of risk.

Neural Predictive Analytical Models

Nt's neural predictive analytical models are developed using specific, rather than generic data: a vital factor as it makes them highly relevant and accurate. An organisation's own historical data, with known outcomes, is used to 'train' the neural computer to learn patterns and characteristics within the data, e.g. age, income, time employed etc. These patterns are then used to create models that can differentiate, for example, between fraud and genuine applications, credit risk or creditworthy applicants, or between those customers likely to remain loyal and those likely to switch to another company.

Neural Behavioural Modelling

As the name suggests, behavioural modelling is use to profile the behaviour of customers. Unique profiles are automatically learned and maintained for each customer, which continually evolve to reflect any changes in normal behaviour - in this way any anomalous and unusual activity is easily recognised.

Nt's behavioural models automatically configure themselves for optimum performance, and hence require no input from a configuration expert other than to specify which features should be profiled. In addition, they offer an adapt-to-feedback facility, which can be used to provide feedback on events that, upon investigation, are found not to pose a risk. This allows the solution to learn from the outcome of investigations. Furthermore, the behavioural models continuously adapt their customer profiles to reflect gradual changes in behaviour over time. This improves performance by guaranteeing that profiles never age and always accurately represent customers' normal behaviour.

Identity Matching Technology

It is common for fraudsters, who are barred from using a service, to slightly change their name and address to prevent the organisation from identifying them when they re-apply for service. Identity matching provides the facility to pinpoint individuals who cannot be identified through traditional subscriber validation techniques.

Link Analysis

Link analysis allows investigators to explore associations between customers by following "links" between the information of different customers. For example, by looking at addresses, family relations, telephone numbers dialled and credit card details it may be possible to identify a larger fraud ring than would be apparent from isolated pieces of information.

Over 30 patents pending & proven neural capability

Nt's solutions support the following operating systems: IBM AIX 5, HP-UX 11i (PA RISC), HP-UX 11i v2 (PA RISC), HP-UX 11i v2 (Itanium), Solaris 8 (SPARC), Solaris 10 (SPARC) and Solaris 10 (X86).