International Business Machines Corporation

Fair Isaac Corporation

Kiyoung Yang

Ramani Ranjan Routray

Prasenjit Sarkar

Eric Kevin Butler

Chung-hao Tan

Scott Michael Zoldi

Yuting Jia

Heming Xu

The current subject matter describes a method and system of detecting frauds or anomalous behavior. The procedures include extracting characteristics from a dataset to generate words and documents, executing a topic model to obtain the respective probabilities of appearance of a document in each latent archetype, dividing the dataset into a plurality of subsets based upon the archetypes. The formed subsets are further utilized to estimate the quantiles and calculate scores using a self-calibrating outlier model. The score of each new transaction is determined based on a single archetype or based on the sum of weighted scores determined from all the archetypes and associated statistics. Such methods are superior to a simple self-calibration outlier model without an LDA archetype. The detection system with the LDA archetypes and self-calibrating outlier model is implemented with the sliding window technique incorporating new transactions into the topic model and it is capable of operating in real-time for the purpose of identifying frauds and outliers.

Real time autonomous archetype outlier analytics

A method and system for automatically identifying an optimal set of attributes of entities included in a networked system. Entity types are ranked based on information gain. A first classification accuracy relative to a first entity type is determined. The first entity type is the top-ranked entity type or a first aggregate entity type. A second entity type is selected base on the ranking. A database join of a first set of attributes associated with the first entity type and a second set of attributes associated with the second entity type is performed. A second classification accuracy relative to a second aggregate entity type generated by the join is determined. In response to determining that the second classification accuracy is not greater than the first classification accuracy, an optimal set of attributes contributing to a problem in the networked system is identified as the first set of attributes.

Automatically identifying an optimal set of attributes to facilitate generating best practices for configuring a networked system

A method and system for automatically identifying an optimal set of attributes of entities included in a networked system. Entity types are ranked based on information gain. A first classification accuracy relative to a first entity type is determined. The first entity type is the top-ranked entity type or a first aggregate entity type. A second entity type is selected based on the ranking. A database join of a first set of attributes associated with the first entity type and a second set of attributes associated with the second entity type is performed. A second classification accuracy relative to a second aggregate entity type generated by the join is determined. In response to determining that the second classification accuracy is not greater than the first classification accuracy, an optimal set of attributes contributing to a problem in the networked system is identified as the first set of attributes.

Growing community of inventors

San Diego, CA, United States of America

Kiyoung Yang