New Research: Machine Learning Classifiers Don't Need Negative Labels

In computer security, researchers usually have easy access only to labels for malicious samples (malware, phishing domains, etc.), while labels for benign samples (productivity software, e-commerce domains, etc.) are missing entirely—or they are tedious and expensive to collect at scale. Typically, this leads to researchers regarding the “known bad” samples as malicious, while the rest is presumed to be benign. DNSFilter's Chief Data Scientist, David Elkind, recently published research that shows that this solution leads to a biased model. In this presentation, David will walk through his findings and explain how DNSFilter’s Malicious Domain Protection uses the concepts in his research to detect threats an average of 10 days sooner than other technologies.