Overview | HNet | Performance Aspects | Mathematics | Biology

A schematic of the cerebellum is presented below.  Afferent (incoming) signals are received from the climbing fiber system from the mid brain, thalamus, and neo-cortex.  Each climbing fiber synapses with between 20 and 40 granule cells.  Granule cells integrate incoming signals, transferring their axonal (output) signals to the dendrites of Purkinje cells.  Synaptic connections between granule axons and the dendritic processes of Purkinje cells are located within the molecular layer.  Purkinje cells have axonal processes which form the only efferent (outgoing) signal from the cerebellum.
Similar to the neo-cortex, granule cells preprocess incoming signals performing a combinatoric expansion through generation of higher order complex scalars.  This expansion provides a very powerful non-linear learning capability to the Purkinje cell.  The cell assembly formed by the combination of granule and Purkinje cells form the basis of the cerebellar model within the HNet Application Development System.
One may use the cerebellar model to construct simple cell assemblies comprised of only four cells.  Despite the simple structure, cerebellar assemblies may perform real time learning over large numbers of stimulus-response patterns.  The cerebellar structure forms the basis for the HNeT SL Platform, and various HNeT applications described on this website.