Machine Learning for Robust Automatic Target Recognition

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. Addressing this challenge, the objectives of this project (Phases I and II) are an investigation and demonstration of prototype algorithms to improve ATR robustness. A common theme underlying robust decision making algorithms is learning from real data which can be accomplished by propagation, interaction, and switching amongst multiple models leading to composite or generalized global models that provide robust and reliable estimates and ATR decisions. We will develop both statistical learning and machine learning based approaches for robust ATR This project will also involve a series of demonstrations, where generalization is conducted over EOC conditions at both the raw data (pixel/intensity) level, and the feature level. Testing at these various levels will account for EOC factors, including clutter and noise variations, minor design differences, pose, revetment, partial obscurations, and articulation of movable parts, etc. The project team consists of Scientific Systems Company, Inc. (SSCI), Woburn MA, and its subcontractor Lockheed Martin Tactical Systems (LMTS), Eagan MN.