Back

Speaker "Siobhan Mcnamara" Details Back

 

Topic

Data Science in Security: Machine Learning for Identity Verification

Abstract

While designing a solution for online identity verification we are dealing with a fast moving target. Those who attempt to deceive are innovative and are incentivized by the potential great rewards from convincing other they are a brand or a person they know or trust. In fact the vast majority of hacks start with an email. People believe they are communicating with someone they know and willingly share sensitive information. In this presentation I will discuss the challenges of both identity verification and of designing models to accurately classify evolving targets.

Profile

I am data scientist working on a security solution at Agari. We have designed a stack of machine learning models that determine if the identity of an e-mail sender is in fact who they purport to be. My background is primarily in research, in economics and psychology. I have a particular interest in risk and uncertainty and the behavioral nuances associated with that. At Agari I have the opportunity to design systems to mitigate risk at scale by measuring the inherent trust and authenticity of online activity.