The Representation of Knowledge in Image Understanding.
Abstract
The author believes that understanding adaptation and knowledge representation is fundamental to make progress in image understanding by animal brains. He has been studying the anatomical structures and the temporal requirements that lead to adaptation. This report makes a brief excursion into some experiments which have been demonstrated in the past. Also discussed is the enormous complexity of the problem in hand and it is concluded that only a method which allows the reading out of recognizable visual memories has any chance to make progress in this complex endeavor. Determined were some of the parameters that produce powerful adaptation and have methods that allow memory read-outs. A preliminary conclusion is that image understanding requires a learning principle that takes into account the nature of the information and not just temporal and/or spatial relationships. Animal brains possess extremely effective vision systems. Architectural and functional principles gained by studying them will certainly lead to new ideas for new computer architectures especially in the fields of machine vision, adaptation, and parallel computation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1985
- Accession Number
- ADA154507
Entities
People
- D. N. Spinelli
Organizations
- University of Massachusetts Amherst