초록 ▼

The invention relates to an apparatus for detecting fraud using a neural network. The architecture of the system involves first employing a conceptual clustering technique to generate a collection of classes from historical data. Neural networks are provided for each class created by the clustering

The invention relates to an apparatus for detecting fraud using a neural network. The architecture of the system involves first employing a conceptual clustering technique to generate a collection of classes from historical data. Neural networks are provided for each class created by the clustering step and the networks are trained using the same historical data. This apparatus is particularly useful for detecting the incidence of fraudulent activity from very large amounts of data such as tax returns or insurance claims.

대표청구항 ▼

[ What is claimed is:] [1.] A method for the classification and detection of complex event spaces using a multi-stage integrated architecture, each event space being represented by a plurality of vectors, said method comprising,selecting a first set of vectors from a first set of data from a databas

[ What is claimed is:] [1.] A method for the classification and detection of complex event spaces using a multi-stage integrated architecture, each event space being represented by a plurality of vectors, said method comprising,selecting a first set of vectors from a first set of data from a database, said first set of data being historical data with known characteristics,providing said first set of selected vectors to a first classification unit, said unit being responsive to said first set of vectors and employing a conjunctive conceptual clustering algorithm, wherein said classification unit provides a plurality of output signals in response to the input vectors each said output signal representing a class,providing neural networks for each said class created by said first classification unit,selecting a second set of vectors from said first set of data for at least one class generated by said first classification unit,training said neural networks with said second set of vectors to reflect a desired output, whereby detection architecture is created, andproviding said detection architecture with a second set of data with unknown characteristics, processing said unknown data with said classification unit by first classifying said data into one of said classes created by said clustering algorithm, and then providing said data to one of said neural networks corresponding to its classification to produce an output signal which reflects a final classification.