Graduation Year
2017
Document Type
Campus Only Senior Thesis
Degree Name
Bachelor of Arts
Department
Computer Science
Reader 1
Robert Keller
Reader 2
Christopher Towse
Rights Information
© 2017 Sneha R Deo
Abstract
Over the past decade, developments in data analysis have improved the quality and efficiency of fraud detection software. Unfortunately, malicious behaviors have also become more subtle, and the quantity of financial data requires techniques to be both accurate and efficient on large-scale and quickly-changing data sets. To address this problem space, the FICO Clinic team was tasked with identifying rare behaviors in unannotated transactional data using topic models. This thesis will describe the development of various novel topic modelling approaches for the detection the latent behaviors that indicate rare events and the methods for their evaluation. Excluded are the results of the project and a description of the novel algorithms, which are the intellectual property of FICO.
Recommended Citation
Deo, Sneha, "Using Latent Topic Models to Detect Rare Events" (2017). Scripps Senior Theses. 1014.
https://scholarship.claremont.edu/scripps_theses/1014
This thesis is restricted to the Claremont Colleges current faculty, students, and staff.
