Exploring Anomalies in Authentication Logs with Autoencoders

Discover how to leverage unsupervised techniques, specifically autoencoders, to detect cybersecurity events in authentication logs in this 33-minute conference talk. Learn about the limitations of traditional rule-based models and explore a more advanced approach that transforms and simplifies complex log data. Gain insights into feeding processed data into an autoencoder and evaluating the output for anomaly detection. Presented by Hayden Beadles, Sr. Security Machine Learning Engineer, and Jericho Cain, Sr Staff Security Data Scientist from Adobe, this talk offers valuable knowledge for enhancing cybersecurity practices using machine learning techniques.

Exploring Anomalies in Authentication Logs with Autoencoders