Abstract: In one embodiment, a device in a network receives sets of traffic flow features from an unsupervised machine learning-based anomaly detector. The sets of traffic flow features are associated with anomaly scores determined by the anomaly detector. The device ranks the sets of traffic flow features based in part on their anomaly scores. The device applies a genetic programming approach to the ranked sets of traffic flow features to generate new sets of traffic flow features. The genetic programming approach uses a fitness function that is based in part on the rankings of the sets of traffic flow features. The device specializes the anomaly detector to emphasize a particular type of anomaly using the new sets of traffic flow features.

Abstract: In one embodiment, a device in a network identifies a plurality of traffic records as anomalous. The device matches each of the plurality of traffic records to one or more anomalies using one or more anomaly graphs. A particular anomaly graph represents hosts in the network as vertices in the graph and communications between hosts as edges in the graph. The device applies one or more ordering rules to the traffic records, to uniquely associate each traffic record to an anomaly in the one or more anomalies. The device sends an anomaly notification for a particular anomaly that is based on the traffic records associated with the particular anomaly.

Abstract: In one embodiment, a device in a network identifies a new interaction between two or more nodes in the network. The device forms a feature vector using contextual information associated with the new interaction between the two or more nodes. The device causes generation of an anomaly detection model for new node interactions using the feature vector. The device uses the anomaly detection model to determine whether a particular node interaction in the network is anomalous.

Abstract: In one embodiment, a device in a network receives feedback regarding an anomaly reporting mechanism used by the device to report network anomalies detected by a plurality of distributed learning agents to a user interface. The device determines an anomaly assessment rate at which a user of the user interface is expected to assess reported anomalies based in part on the feedback. The device receives an anomaly notification regarding a particular anomaly detected by a particular one of the distributed learning agents. The device reports, via the anomaly reporting mechanism, the particular anomaly to the user interface based on the determined anomaly assessment rate.

Abstract: In one embodiment, a device in a network maintains a plurality of anomaly detection models for different sets of aggregated traffic data regarding traffic in the network. The device determines a measure of confidence in a particular one of the anomaly detection models that evaluates a particular set of aggregated traffic data. The device dynamically replaces the particular anomaly detection model with a second anomaly detection model configured to evaluate the particular set of aggregated traffic data and has a different model capacity than that of the particular anomaly detection model. The device provides an anomaly event notification to a supervisory controller based on a combined output of the second anomaly detection model and of one or more of the anomaly detection models in the plurality of anomaly detection models.

Abstract: In one embodiment, a device in a network determines cluster assignments that assign traffic data regarding traffic in the network to activity level clusters based on one or more measures of traffic activity in the traffic data. The device uses the cluster assignments to predict seasonal activity for a particular subset of the traffic in the network. The device determines an activity level for new traffic data regarding the particular subset of traffic in the network. The device detects a network anomaly by comparing the activity level for the new traffic data to the predicted seasonal activity.

Abstract: In one embodiment, a device in a network receives traffic metrics for a plurality of applications in the network. The device populates a feature space for a machine learning-based anomaly detector. The device identifies a missing dataset in the feature space for a particular one of the plurality of applications. The device adjusts how traffic is sent in the network, to capture the missing dataset.