Date: | 2020, June 16 |
Time: | 11:00 |
Place: | Online |
Author: | Mossakowski, Till (Moderator) |
Title: | Copycat – Creative Fluid Analogies (A Discussion) |
These days, limits of deep learning become more and more appearant, motivating the need for subsymbolic-symbolic or neural-symbolic integration, being very roughly analogous to the integration of left and right brain hemisphere in humans. Pioneering work in the area of subsymbolic-symbolic integration is Copycat, a cognitive architecture for creating creative fluid analogies for a very simple domain of short strings. This architecture has also been used for other domains: there is Phaeaco (a Bongard problem solver), SeqSee (number sequence extrapolation), George (geometric exploration), and Musicat (a melodic expectation model). The attached chapter will enable use to have a look at both some sample analogies from the string domain as well as at the architecture, and discuss possible inspiration for current research on subsymbolic-symbolic integration.