Abstract
BOSS is a highly parameterized large-scale discrete- event simulation system for brain structures. BOSS has been written in FORTRAN for the Univac 1106 and in PASCAL for the Control Data Cyber 173. It permits the simulation of detailed three-dimensional models involving up to 8,000 neurons and 80,000 synapses. Instead of describing the individual elements of a model, parameters specify spatial lattices and intersecting probability distributions which position neurons and synapses with any desired variance. To avoid sorting overhead, time is finely quantized and events are posted and scanned via a circular array of pointers to lists of simultaneous events. Synap tic weights may be modified systematically to reflect changes resulting from learning. BOSS can be used to simulate large logic arrays as well as neural networks.
Keywords
Get full access to this article
View all access options for this article.
