The problem of how abstract symbols, such as those in systems of natural language, may be grounded in perceptual information presents a significant challenge to several areas of research. This thesis presents an unsupervised learning model that allows analysis of the symbol-grounding problem. The model learns associations between visual scenes and linguistic descriptions and provides means for direct examination of what it has learned. By analyzing the system, it is possible to assess how well symbols can be grounded in perceptual information with an unsupervised neural network architecture. The model demonstrates potential for accomplishing grounding in artificial systems and provides valuable insight into the grounding task.

PDF

Technical Report TR-06-01, Department of Computer Science, The University of Texas at Austin, 2005.

@techreport{williams:ugthesis05, title={Learning Visual Scene Descriptions: An Approach to Symbol Grounding}, author={Paul Williams}, number={TR-06-01}, institution={Department of Computer Science, The University of Texas at Austin}, type={Undergraduate Honors Thesis}, url="http://nn.cs.utexas.edu/?williams:ugthesis05", year={2005} }