Computational synthesis tools that automatically generate solutions to design problems are not widely used in industry despite many years of research. This deficiency can be attributed to the lack of value that these tools provide for the user in terms of time saved in design or quality improvements in the design. In order to provide sufficient quality of solution, it is proposed that more human-like evaluation of solution quality is needed including qualitative concepts, the ability to allow for anecdotal input, and general inclusion of ambiguous information. A hierarchical temporal memory system (HTM) is proposed as a viable approach for capturing design quality from exemplars and subsequently recognizing the presence of that quality in other designs. This paper includes a first experiment in using HTMs for learning and recognizing quality in the form of the visual style characteristics of Hepplewhite, Stickley, and Greene & Greene chair backs. Results show that HTMs develop a similar storage of quality to humans and are therefore a promising option for capturing and recognizing multi-modal quality information in future design automation projects.

This content is only available via PDF.
You do not currently have access to this content.