Breakthrough Research: Machine Learning Success Without Negative Labels
John, in the realm of computer security, researchers face a persistent challenge: they typically have abundant access to labeled malicious samples (malware, phishing domains, etc.) but struggle to obtain comprehensive labels for benign samples (legitimate software, trustworthy websites, etc.). Collecting these benign labels often proves prohibitively time-consuming and expensive at scale.
This imbalance has led to a common but flawed approach—classifying known samples as malicious while presuming everything else to be benign.
DNSFilter's Chief Data Scientist, David Elkind, has published groundbreaking research demonstrating how this conventional approach creates inherently biased security models. In this upcoming webinar, David will share his innovative findings and demonstrate how DNSFilter's Malicious Domain Protection leverages these research insights to identify threats an average of 10 days faster than competing technologies. |
|
|
|
| Can't attend the event? Register anyway and we will let you know when it is available on-demand. |
|
|
|
|
