A 10 Networks, Incorporated

Other

Vishnu Vasanth Radja

Vernon Richard Groves

Diptanshu Purwar

Micheal Thompson

Madhav Aggarwal

A method and system for detecting malicious network packets via an intelligent large language model are described. In addition, a method for fine-tuning a pretrained large language model to detect malicious network packets is described. The training process generates a plurality of embeddings from input network packet data, generates clusters for those embeddings, performs an entropy analysis, and calculates a loss value. Contrastive learning is used to further fine-tune the large language model based embedder.

Large language model based intelligent malicious packet detection

Provided are methods and systems for cluster-based mitigation of a network attack. A method for cluster-based mitigation of a network attack may commence with detecting an unusual pattern in network data traffic associated with data sources. The method may further include extracting signature parameters associated with the network data traffic. The signature parameters may be indicative of the network attack. The method may continue with assigning importance weights to the signature parameters based on historical signature data to generate weighted signature parameters. The method may further include building a decision tree for the data sources based on the weighted signature parameters. The method may continue with creating an optimal number of clusters for the data sources based on an analysis of the decision tree. The method may further include selectively taking at least one mitigating action with regard to the data sources within the clusters.

Cluster-based precision mitigation of network attacks

Provided are methods and systems for cluster-based determination of signatures for detection of anomalous data traffic. An example method may include capturing, by a network module, data packets routed to a destination. The method may further include grouping, by at least one processor in communication with the network module, the data packets into clusters. The method may also include detecting, by the processor, an anomaly in the data packets and, in response to the detection, determining, by the processor and based on the clusters, one or more signatures associated with the data packets. The method may further include generating, by the processor and based on the signatures, one or more rules for allowing the data packets. The method may further include providing, by the processor, the one or more rules to a policy enforcement point associated with the destination.

Growing community of inventors

San Jose, CA, United States of America

Vishnu Vasanth Radja