Detection and Investigation of Email Exfiltration Events in Sun Life Cybersecurity Data, including Text Analysis

The current Sun Life anomaly detection methods are based on human inspection. The method is labour intensive and only a relatively small number of cases can be reviewed. There is no guarantee that the most suspicious interactions are in fact being sampled and reviewed. In this project we will focus on the develop of novel AL methods and user interfaces that can improve the process of labelling instances and recognizing if anomalies are likely to be malicious, and we improve the sensitive of automated anomaly detection models by adding text analysis into the process. Our work will involve cutting edge research in the emerging field of human-AI interaction, and improving the quality of detection our research to: a) Reduce workload for experts involved in the detection process; b) Improve trust/compatibility of the expert-model team; c) Develop new insights concerning the frequency and types of risky behaviours that are carried out by employees.

Faculty Supervisor:

Mark Chignell

Student:

Partner:

Sun Life Financial

Discipline:

Engineering

Sector:

Finance and Insurance

University:

University of Toronto

Program: