Vectra AI has secured a grant for a innovative system that employs machine learning to identify malicious payloads within networks. This cutting-edge technology aims to enhance cybersecurity measures by accurately detecting potential threats in network traffic.
The recently granted patent, with Publication Number US11973768B2, highlights a sophisticated system that utilizes communication over distinct server and port combinations in a network setting. The system includes a metadata processing element for extracting metadata from client-server sessions, a learning module for creating behavior models for various server-port combinations, and an update module for refining these models based on new communications. By calculating a similarity score using a hamming distance between baseline patterns, the system can effectively evaluate the relevance of new communications to existing behavior models.
Additionally, the patent describes a method and computer-readable medium that operate similarly to the system by extracting metadata from network traffic, generating behavior models for server-port combinations, and updating these models with new communications. A detection module is included in the method to trigger alerts when a client-server session deviates from the expected behavior model, potentially indicating malicious activity. Overall, the patent underscores the application of machine learning techniques to analyze network traffic and detect anomalies based on behavior models associated with specific server and port combinations within a network environment.
