ML Anomaly Detection Training: Start With the Base Rate
A developer's guide to training ML anomaly detection systems argues that evaluation must start with the base rate of attacks, citing Los Alamos National Laboratory data where malicious authentication …