Machine Learning for Robust Automatic Target Recognition
Agency / Branch:
DOD / USAF
The primary issue faced by ATR systems center around the large "variability of the ATR problem space spanned by the targets, sensors, and the environment and the associated challenge is to develop "robust" approaches to improve ATR system performance. The primary innovation of this work is the development of an automated way of developing heuristic inference rules that can draw on multiple models and multiple feature types to make more robust ATR decisions. The key realization is that this ``meta learning' problem is one of structural learning, that can be conducted independently of parameter learning associated with each model and feature based technique, and more effectively draw on the strengths of all such techniques, and even information from unforeseen techniques. We will accomplish this by using robust, genetics-based machine learning for the ill conditioned combinatorial problem of structural rule learning, while using statistical and mathematical techniques for parameter learning. This project will also involve a series of demonstrations, where generalization is conducted over EOC conditions. Testing at these various levels will account for a variety of EOC factors, including clutter and noise variations, minor design differences, pose, revetment, partial obscurations, and articulation of movable parts, etc.Success will be measured in several ways (each as compared to standard ATR techniques). The project team consists of Scientific Systems Company, Inc. (SSCI), Woburn MA, and its subcontractor Lockheed Martin Tactical Systems (LMTS), Eagan MN.
Small Business Information at Submission:
Manager R & D
President and CEO
SCIENTIFIC SYSTEMS CO., INC.
500 West Cummings Park - Ste 3000 Woburn, MA 01801
Number of Employees: