US Patent 11025665 Detection and identification of targeted attacks on a computing system

Malicious activity data is obtained, that is indicative of attempted attacks on a computing system. Clusters of targets are identified and it is determined whether the malicious activity preferentially targets one cluster of targets over other. Also, low prevalence attacks are identified and it is determined whether a low prevalence attack has a high concentration in one or more of the target clusters. If the malicious activity either preferentially targets a cluster, or a low prevalence attack has a high concentration in a cluster, then the attack is identified as a targeted attack, so that remediation steps can be taken.