Your Training Data is Bad and You Should Feel Bad

DerbyCon 8.0 - Evolution

Presented by: Ryan J. O'Grady
Date: Saturday October 06, 2018
Time: 17:00 - 17:25
Location: Kentucky C & D
Track: Stable

Everyone is using Big Data and Machine Learning these days. Not sure how to solve a problem? Train a classifier! But beware the old axiom: Garbage In, Garbage Out. This talk will present three key findings from original research on the effects of training data recency in Twitter classifiers so that your next Twitter bot classifier can start off on the right footing.

Ryan J. O'Grady

Ryan O’Grady has worked in cyber security for over 10 years and is a research scientist in Soar Technology’s Cyberspace Operations business area. He is the principal investigator for a project to develop an intelligent training system for cyberspace operators that enables individualized, personalized training in realistic environments. He has a BSE in Computer Science from the University of Michigan and is currently pursuing a MS in Information Security Engineering from SANS Technology Institute


KhanFu - Mobile schedules for INFOSEC conferences.
Mobile interface | Alternate Formats