Symbolic and SubsymbolicLearning for Vision: SomePossibilities

Robust, flexible and sufficiently general vision systems such as those for recognition and description of complex3-dimensional objects require an adequate armamentarium of representations and learning mechanisms. This paper briefly analyzes the strengths and weaknessesof different learning paradigms such as symbol processing systems, connectionist networks, and statistical and syntactic pattern recognition systems as possible candidates for providing such capabilities and points out several promising directions for integrating multiple such paradigms in a synergistic fashion towards that goal.