TIGP (SNHCC) -- Fighting Abuse on Facebook with Machine Learning
- LecturerDr. Hung-Pin Kao (Facebook)
Host: TIGP SNHCC Program - Time2016-02-17 (Wed.) 14:00 ~ 16:00
- LocationAuditorium 106 at IIS new Building
Abstract
Like any popular Internet site, Facebook is a target for abuse. Facebook engineers heavily rely on machine learning in fighting and mitigating the negative effects. We use classification and clustering in operations from catching fake accounts, detecting violating photos, to filtering spam comments. This presentation will focus on the machine learning life cycle in Facebook and the related infrastructure. It will cover selected abuses and show how Facebook developed and applied models in protecting its community.
BIO
Hung-Pin Kao received his Ph.D. in Quantitative Methods from Purdue University in 2010. After four years with Amazon, he joined Facebook in 2014. His interests are in machine learning, big data analytics and high-availability-low-latency systems. He is currently with the machine learning team in Facebook’s Care Engineering.