Description:
Integrated Prudence Analysis has been proposed as a method to maximize the accuracy of rule based systems. The paper presents evaluation results of the three Prudence methods on public datasets which demonstrate that combining attribute-based and structural Prudence produces a net improvement in Prudence Accuracy.
Description:
The threat posed by cybercrime to individuals, banks and other online financial service providers is real and serious. Through phishing, unsuspecting victims’ Internet banking usernames and passwords are stolen and their accounts robbed. In addressing this issue, commercial banks and other financial institutions use a generically similar approach in their Internet banking fraud detection systems. This common approach involves the use of a rule-based system combined with an Artificial Neural Network (ANN). The approach used by commercial banks has limitations that affect their efficiency in curbing new fraudulent transactions. Firstly, the banks’ security systems are focused on preventing unauthorized entry and have no way of conclusively detecting an imposter using stolen credentials. Also, updating these systems is slow and their maintenance is labour-intensive and ultimately costly to the business. A major limitation of these rule-bases is brittleness; an inability to recognise the limits of their knowledge. To address the limitations highlighted above, this thesis proposes, develops and evaluates a new system for use in Internet banking fraud detection using Prudence Analysis, a technique through which a system can detect when its knowledge is insufficient for a given case. Specifically, the thesis proposes the following contributions:
Description:
The threat posed by cybercrime to individuals, banks and other online financial service providers is real and serious. Through phishing, unsuspecting victims’ Internet banking usernames and passwords are stolen and their accounts robbed. In addressing this issue, commercial banks and other financial institutions use a generically similar approach in their Internet banking fraud detection systems. This common approach involves the use of a rule-based system combined with an Artificial Neural Network (ANN). The approach used by commercial banks has limitations that affect their efficiency in curbing new fraudulent transactions. Firstly, the banks’ security systems are focused on preventing unauthorized entry and have no way of conclusively detecting an imposter using stolen credentials. Also, updating these systems is slow and their maintenance is labour-intensive and ultimately costly to the business. A major limitation of these rule-bases is brittleness; an inability to recognise the limits of their knowledge. To address the limitations highlighted above, this thesis proposes, develops and evaluates a new system for use in Internet banking fraud detection using Prudence Analysis, a technique through which a system can detect when its knowledge is insufficient for a given case. Specifically, the thesis proposes the following contributions:
Description:
Rated Multiple Classification Ripple Down Rules (RM) and Ripple Down Models (RDM) are two of the successful prudent RDR approaches published. To date, there has not been a published, dedicated comparison of the two. This paper presents a systematic preliminary evaluation and analysis of the two techniques. The tests and results reported in this paper are the first phase of direct evaluations of RM and RDM against each other.