Learning Maneuvers Using Neural Network Models.

Abstract : This grant covered the completion of the PhD thesis of Paul Viola and the initiation of the PhD work of Oded Maron. Viola's work was on alignment of 2 and 3 dimensional objects based on maximization of mutual information. The technique depends only on object shape and is robust to variations of illumination. The algorithms are quite general and can foreseeably be used in a wide variety of imaging situations. Maron's work has focused on a variation on supervised learning called multiple-instance learning, where the task is to learn a concept given positive and negative bags of instances. Each bag may contain many instances but a bag is labeled positive even if only one of the instances in it falls within the concept. A bag is labeled negative only if all the instances in it are negative. This framework has been applied to a variety of problem domains.