CORDIS provides links to public deliverables and publications of HORIZON projects.
Links to deliverables and publications from FP7 projects, as well as links to some specific result types such as dataset and software, are dynamically retrieved from OpenAIRE .
Exploitable results
Two system prototypes for fraud detection have been developed in Credit Cards and in Telecommunication sectors.
The COLBA project aimed to develop system prototypes for fraud detection. Two specific neural prototypes have been developed and a set of analyses and experiments has been performed on high performance systems. In greater detail, the following results have been generated: - a market study in the credit card, telecommunication and electronic service markets; - a neural fraud detection prototype for credit card customers; - a neural fraud detection software prototype for merchants; - a representative database implementing behavioural data on customers and merchants; - a feasibility study on the extension of the high performance neural fraud detection technology to the telecommunication sector; - an analysis of the performance of the credit fraud detection prototypes and databases on SMP machines. The performance figures have also been analysed with respect to the telecommunications sector requirements.
Different types of results have been obtained. In the Telecom sector, the neural-network-based system has been designed to operate in conjunction with conventional rule-based systems. The system uses a number of neural network models to characterise the normal behaviour of the customer base and to detect when deviations occur. Alerts are generated based on the probability of fraud for a given account. Fraud investigation staff can then investigate these alerts. In the Credit cards area, the study has made clear that the neural technology is some way redundant for Italian credit cards customers spending profiles management. Simpler, rule-based systems can be more suitable (and cheaper) to the frauds detection purpose. It is expected that using the developed prototype a large frauds reduction can occur, in the range of 30-40% of the actual costs. It is foreseen that an on-line system could be developed in about one year.
Applications of COLBA (Credit cards) results are directly related to the integration of the fraud detections models with existing authorisation systems. In detail, this means the implementation of an on-line fraud detection system able to generate (alert) signal at each incoming transaction. Condition sine qua non is the availability of a database of customers/merchants profiles containing information on established behaviours. For this reason applications of the results could also be oriented to market analysis for marketing purposes. The COLBA project performed a feasibility study and has verified that the neural network approach in the Telecoms area can work using a simulation in an off-line environment.
Searching for OpenAIRE data...
There was an error trying to search data from OpenAIRE
No results available