Iterative Sub-Setting

摘要

The objective of this document is to communicate the progress made on an initiative to improve the current Case-Based Reasoning (CBR) functionality of the Multi-Intelligence Tools Suite (MITS) platform. To do so, a new similarity measure based on information theory called Iterative Sub- Setting (ISS) is proposed. The goal of this initiative was to provide a potential solution to a list of issues identified in the literature about Maritime Anomaly Detection (MAD). Various tests with this measure were made to demonstrate the potential for it to address selected issues in MAD. Results show that this method is more sensible to the context where the CBR is performed, can be used with a small case base, evolve over time and provide an accurate confidence level on the selected cases.