Collaborative Learning Agent (CLA) For Trident Warrior 2008

Description

Port security is important. The Navy needs to enhance its awareness of potential threats in this dynamic environment—and plan for potential high-risk events such as use of maritime shipping for WMD or other malicious activities.  The CLA technology is used to learn knowledge patterns from historical Maritime Domain Awareness (MDA) data and then uses the patterns for prediction and identification of anomalies and reasons that might cause the anomalies, e.g. weather or potential terrorist activities.  A single CLA learns and discovers knowledge and behavior patterns from a single historical data source and the patterns are used for behavior prediction for a new data.  A set of networked Collaborative Learning Agents (CLAs) forms an agent network and performs a collaboration to enhance the prediction.

We use three agents in this exercise that can learn patterns from historical maritime domain information.  Agent 1 (http://cla1.quantumii.com/FAIRPLAY) learns patterns from The Lloyd’s Register – Fairplay (LRF) news; Agent 2 (http://cla2.quantumii.com/JOC) from the Journal of Commerce, which includes information regarding port events, activities, rules, and policies; and Agent 3 (http://cla3.quantumii.com/MPC) from Maritime Press Clippings which are freelance vessel and incident reports.  The historical data is up to 6/15/2008. Users need only to observe the real-time test process, as shown below, as opposed to the above learning process.

In a real-time test process, when a piece of real-time information is newly observed, it goes through the CLA network; the network then returns a report of search results which shows if the new information is correlated with the patterns and to what degree the correlation is.  In this exercise, the real-time data from the SPAWAR MDA DS COI (https://mda.spawar.navy.mil), including a vessel’s name and position, is collected from each vessel’s Automatic Identification System (AIS) data.  Each real-time input goes through the CLA network and is classified into prediction categories (see Figure 1): 1) Anomaly (red), i.e. a search input that has low correlation with previously discovered context patterns; 2) Relevant (green), i.e. an input is highly correlated to the previously discovered knowledge patterns; 3) Medium Correlation (yellow), i.e. between relevant and anomaly; 4) Irrelevant (white), i.e. an input is not related to any of the agents’ knowledge patterns, or a correlation value can not be computed from the CLA network.

Figure 1: Anomaly Meter

A user will observe the test process for about 100 real-time inputs.  Each input (sequence) represents a vessel’s name or real-time location from the SPAWAR MDA DS COI.  The input is checked against the patterns in the CLA network to see if anything is of interest or relevance to the vessel or its location; for example, was the vessel seen anywhere else before? Were there any incidents/activities/events reported in the vessel’s location? A user will compare samples of the categorizations (i.e. anomaly, relevant, medium correlation or irrelevant) from the CLA network with his/her own categorization.