Kaspersky Machine Learning for Anomaly Detection 5.0 has the following new capabilities and improvements:
Model builder: added support for elliptic envelope-based ML models. Added functionality that lets you log incidents when observing anomalous behavior for a certain interval of time. Added functionality that lets you clone elements of ML models and markups. Added display of statuses and states of ML models and their elements in the asset tree. Added functionality that lets you search and filter by elements of the asset tree.
Monitoring and History sections: added support for displaying charts of several tags within the same graphic area. Graphic areas and the tags within them are managed in the Presets section.
Starting and stopping the application: added functionality that lets you start and stop the application by using the systemctl utility.
User accounts: added capability to change email addresses for user accounts in the web interface.
Incident notifications: added capability to specify additional email addresses for sending incident notifications in the web interface. Added functionality that lets you send notifications when registering certain types of incidents: incidents registered by certain types of ML model elements or registered by the Limit Detector and/or Stream Processor service. Added functionality that lets you prevent forwarding of notifications about incidents registered by elements of unpublished ML models.
Event Processor service: expanded functionality of the attention mechanism. Added functionality that lets you track generalized events and patterns. Improved display of monitors.
MQTT Connectors, AMQP Connector and CEF Connector - added support for TLS connections by default, and added functionality that lets you use the recommended TLS connection settings.
WebSocket Connector: added functionality that lets you use the recommended TLS connection settings.
Push messages: added functionality that lets you transfer messages between Kaspersky MLAD services.