COGS 100 Lecture Notes - Lecture 15: Edge Detection, Computational Neuroscience, Connectionism

• School: Simon Fraser University
• Department: Cognitive Science
• Course: COGS 100
• Professor: Mark Blair

Document Summary

David marr (1945-1980: one of the founders of computational neuroscience, look for explanations of phenomena on 3 levels, explaining the calculator. What are the precise info processing steps that is happening: steps carried out by calculator, its program. E. g. shades of white-gray-black: high number if bright, low number if dark, plotting the graph in numeric representation. Images are intrinsically noisy (lots of random variation) Images are organized at different special scales (both small lines and edges and also large ones: e. g. so(cid:373)eo(cid:374)e"s hair, (cid:271)lot(cid:272)hes o(cid:374) (cid:272)hairs. Images must be processed quickly: there might be danger in the world. E. g. a center-surround field (a) and its response profile (b) 2d dog filter. Marr"s u(cid:374)dersta(cid:374)di(cid:374)g (cid:373)ultiple levels a(cid:374)d (cid:373)appi(cid:374)g the levels o(cid:374)to ea(cid:272)h other. Potential advantages: e. g. one part of the brain is too active. Therefore invent a drug to make less active parts, more active. Choosing the algorithm: ruling out inconsistent ones.