Network Packet Capture Analysis Using Machine Learning Model

A method of training machine learning models for performing network packet analysis, comprising:

• extracting a subset of network packets from network packets received from a network;
• correlating network packets in the subset of the network packets;
• tagging the correlated network packets with first labels that identify successes or failures in subtasks of communication flows associated with the correlated network packets;
• tagging the correlated network packets with second labels that identify successes or failures of the communication flows associated with the correlated network packets;
• extracting first features from at least one of the first labels or the second labels;
• and
• training a first machine learning model using the extracted first features to predict or infer successes or failures of subsequent communication flows.

A network traffic analysis device, comprising:

• a processor;
• and
• memory storing instructions thereon, the instructions when executed by a processor cause the processor to:
• receive network packets from a location in a network,
• preprocess the received network packets to extract a subset of the network packets, correlate network packets in the subset of the network packets, and tagging the correlated network packets with labels that identify successes or failures in a subtask of a communication flow associated with the correlated network packets,
• extract first features from the correlated network packets and the labels,
• apply the first features to a first machine learning model to predict or infer a success or a failure of the communication flow associated with the correlated network packets, and
• generate an inference output based on the prediction or inference.