| id |
20ff |
| authors |
Derix, Christian |
| year |
2004 |
| title |
Building a Synthetic Cognizer |
| source |
Design Computation Cognition conference 2004, MIT |
| summary |
Understanding ‘space’ as a structured and dynamic system can provide us with insight into the central concept in the architectural discourse that so far has proven to withstand
theoretical framing (McLuhan 1964). The basis for this theoretical assumption is that space is not a void left by solid matter but instead an emergent quality of action and interaction between individuals and groups with a physical environment (Hillier 1996). In this way it can be described as a parallel distributed system, a self-organising entity.
Extrapolating from Luhmann’s theory of social systems (Luhmann 1984), a spatial system is autonomous from its progenitors, people, but remains intangible to a human observer due to its abstract nature and therefore has to be analysed by computed entities, synthetic cognisers, with the capacity to perceive.
This poster shows an attempt to use another complex system, a distributed connected algorithm based on Kohonen’s self-organising feature maps – SOM (Kohonen 1997), as a
“perceptual aid” for creating geometric mappings of these spatial systems that will shed light on our understanding of space by not representing space through our usual mechanics but by constructing artificial spatial cognisers with abilities to make spatial representations of their own. This allows us to be shown novel representations that can help us to see new differences and similarities in spatial configurations. |
| keywords |
architectural design, neural networks, cognition, representation |
| series |
other |
| type |
poster |
| email |
|
| more |
http://www.springer.com/computer/ai/book/978-1-4020-2392-7 |
| full text |
 file.pdf (2,382,336 bytes) |
| references |
Content-type: text/plain
|
| last changed |
2012/09/17 21:13 |
|
