Neurocomputers With Dynamic Connectivity
Provisional Patent Application for Letters
Patent of the Inited States,
Serial No. 60/108,353.
Frank C. Hoppensteadt and Eugene M. Izhikevich
Systems Science Center, Box 7606,
Arizona State University,
Tempe, AZ 85287-7606.
Abstract. We suggest
a novel design of neuro-computers using the following principles:
When the input is constant or has inappropriate frequencies,
the oscillators do not interact due to the same principle used in selective
tuning in radio. However, the rhythmic input can force any two oscillators
to interact, which we call dynamic connectivity, if the input’s Power spectrum
includes the frequency equal to the difference between the frequencies
of the oscillators. Thus, the strength and arrangement of connections between
the oscillators is not hard-wired, but it is dynamic and can be controlled
by the rhythmic input.
Each neuron in the neuro-computer is a high-frequency
The oscillators have distinct frequencies.
The oscillators are weakly connected by a common
The common medium is forced by an external rhythmic
There can be many realizations of such oscillatory
neuro-computers since there are many mechanisms for producing these oscillatory
behaviors. For example, the oscillators could be phase locked loops, VCOs
(Voltage Controlled Oscillators), MEMS resonators, Josephson junctions,
lasers, macromolecules, etc.
Detailed information is available.