Processing data flows with a data flow processor

A network apparatus for processing data flows, comprising:

• a chassis;
• one or more memories within the chassis;
• and
• one or more network processors within the chassis, the one or more network processors configured to execute instructions stored in the one or more memories to:
• receive and forward a stream of data packets in a network;
• recognize one or more data packets in the stream of data packets that contain data, including subscriber profile information, to be processed by an application executing on the network apparatus by applying a policy to the data;
• define an application suite by storing a plurality of applications in the one or more memories including at least two of:
• a virus detection application, an intrusion detection application, a firewall application, a content filtering application, a privacy protection application, and a policy-based browsing application;
• select an application of the plurality of applications stored in the one or more memories for processing the stream of data packets based on payloads of the data packets and on the subscriber profile information
• execute the selected application so as to process the stream of data packets by applying the policy to the payloads using machine learning logic to dynamically reconfigure a data flow, resulting in processed data, the machine learning logic configured to:
• compare a feature vector of the data flow with each of a plurality of artificial neurons that populate an array with each of the plurality of artificial neurons characterized by a weight vector;
• declare the weight vector positioned at the smallest Euclidean distance from the feature vector to be the winning neuron;
• map the feature vector to the winning neuron;
• repeat the comparing, declaring, and mapping with additional feature vectors to create an output map;
• determine whether the data flow is anomalous by determining whether the output map is atypical due to at least one value in the output map being larger or smaller than a threshold in relation to other values in the output map;
• and
• return the processed data for forwarding to a destination in the network.