7 Neural Networks
Smith 1996-2001
On the previous page is a diagram of a typical neural network called a Back-Propagated Delta Rule Network. We can read below what a neural network implies:
Neural Networks are a different paradigm for computing:
* von Neumann machines are based on the processing/memory abstraction of human information processing.
* neural networks are based on the parallel architecture of animal [human] brains.
Neural networks are a form of multiprocessor computer systems, with
* simple processing elements
* a high degree of interconnection
* simple scalar messages
* adaptive interaction between elements
(Smith 2001:1)
As we read in the explanation above, neural networks are based on the processing / memory abstraction of human information processing. A cognitive scientific approach of (artisticly) mapping the computer data to sound is recently being researched. Scientists try to find a correlation between neural network computer programming concepts and the physicality of the perception of sound composition as we can read below:
There are discrete brain systems and computations for particular musical experiences and skills, these systems are distributed through the whole brain (i.e., left and right cerebral cortex, sub-cortex, and cerebellum). Distinct distributed brain systems serve different musical features such as timbre, harmony, melody, meter, tempo, dissonance, and consonance. Different musical parameters are served by distinct neural systems and networks … Evidence suggests that music is a consequence of biological evolution and is therefore associated with a specific brain architecture (Parsons 2003).
Connectivity Connectionism is the way of cognitive scientists, connectivism is the way of the technoetic artist. They converge where the artificial collaborates with the natural in a new synthesis of being (Roy Ascott 1996)
Aesthetics The classical concern with the surface image of the world gives way to the technoetic aesthetics of creative consciousness and artificial life. (Roy Ascott 1996)
Tod Winkler makes us aware though that, however interesting this sound programming concept is, it is important to realise that the artist should still have the power and freedom to shape the sound in a subjective manner:
An examination of the physical parameters of movement, their limitations, and the methods used for their measurement, will yield clues to an appropriate musical response, one where imaginative connections engage performers by their power to shape sound. Interactive music systems can be used to interpret these data, extending the performer's power of expression beyond a simple one-to-one relationship of triggered sound, to include the control of compositional processes, musical structure, signal processing, and sound synthesis
(Winkler 1995:1).
Noetic networks Our personal neural networks merge with global networks to create a new space of consciousness (Roy Ascott 1996)
Parsons, L. M. (2003) Brain Basis of Musical Performance, Cognition, Perception, and Improvisation Subtle Technologies Conference. Cognitive Neuroscience, National Science Foundation, Washington, DC Research Imaging Center, University of Texas Health Science Center.
Smith, L. (1996-2001) An Introduction to Neural Networks IEEE Neural Networks Society UKRI (United Kingdom and Republic of Ireland).
Winkler, T. (1995) Making Motion Musical: Gesture Mapping Strategies for Interactive Computer Music. Proceedings of the 1995 International Computer Music Conference, Banff, Canada.