Symbolic Visual Learning

Some of the fundamental constraints of automated machine vision have been the inability automatically to adapt parameter settings or utilize previous adaptations in changing environments. Symbolic Visual Learning, as presented in this book, consists of an area of research that tries to overcome these fundamental constraints, enhancing state-of-the-art recognition systems that can measure their own performance, learn from their experience, and outperform conventional static designs. It was written as a companion volume to Early Visual Learning edited by S. Nayar and T. Poggio.