A FortiWeb administrator wants to create a machine learning (ML)-based bot detection system.

FortiWeb machine learning protection depends on observed application traffic. The administrator must first collect traffic samples so FortiWeb can learn normal behavior and create a useful baseline. After sample collection, FortiWeb uses the collected data to build the detection model. Once the model is built, it must be enabled or run in the live environment so FortiWeb can evaluate production requests and detect abnormal or bot-like behavior. Manual verification on test data only is not enough to activate the model for real traffic. Bayesian analysis is not the FortiWeb configuration step shown for this process; the platform handles model logic internally. The practical workflow is collection, model building, and live enforcement or detection.

================