Associative network and signal handling element therefor for pro

Electrical transmission or interconnection systems – Nonlinear reactor systems – Parametrons

Patent

Rate now

  [ 0.00 ] – not rated yet Voters 0   Comments 0

Details

364807, 395 24, G06F 1542

Patent

active

050721305

DESCRIPTION:

BRIEF SUMMARY
This invention relates to signal processing, and especially, but not exclusively to pattern recognition and especially to the relaxation and recognition of patterns in digital or analog signals.
Electronic information stores such as Random Access Memories (RAMs), Read Only Memories (ROMs), Programmable ROMs (PROMs) and Electrically Erasable PROMs (EEPROMs), are known in which target information is stored by disconnecting the links of a lattice structure which otherwise connect a series of column conductors with a series of row conductors lying across the column conductors. To access the information thus stored in the memory the rows have to be addresed vertically, and the corresponding output of the columns read according to the pattern of made and broken links. However RAMs, ROMs, PROMs and EEPROMs do not take into account the possibility of noisy or erroneous addressing, and have no facility with which to correct such an inaccurate input address vector.
A paper entitled "A novel associative memory implemented using collective computation" by M. Sivilotti, M. Emerling, and C. Mead at pages 329 to 342 of Proceedings of the Chapel Hill Conference on VLSI, 1985, edited by H. Fuches and published in 1986 by Computer Science Press, contains a discussion of the possibility of a signal processing network that, in response to a `key` input pattern, provides an output pattern which `relaxes` with time towards a stable state representing the stored `target pattern` most similar to, or the best fit' to the key pattern. A complex network circuit is also described.
It is an object of the present invention is to provide a circuit capable of a best fit so-called auto-associative memory function but based on simpler network elements. More generally, the invention is intended to perform relaxation or `constraint satisfaction` tasks under which key patterns fed into the network are translated into a form consistent with rules and/or constraints programmed into the connectivity of the network by either the manufacturer or the end-user.
According to a first aspect of the invention, there is provided a signal handling element comprising a first signal port, a second signal port, accumulating means and adjusting means, the accumulating means being capable of accumulating signals supplied to the first signal port and the adjusting means being capable of reducing the contents of the accumulating means over a period of time, and means for supplying an output signal to the second port, the arrangement being such that when the contents of the accumulating means are above a certain value, an output signal is not supplied to the second port.
The output signal supply means can be embodied by a signal generator whose operation or output is inhibited when the contents of the accumulating means satisfy the given condition.
The adjusting means may be a dissipator.
According to a second aspect of the invention, there is provided a network comprising a plurality of elements, each element including a first signal port, a second signal port, accumulating means and adjusting means, the accumulating means being capable of accumulating signals supplied to the first signal port, and the adjusting means being capable of altering the contents of the accumulating means over a period of time; and means for supplying an output signal to the second port, the arrangement being such that when the contents of the accumulating means satisfy a given condition, an output signal is not supplied to the second port; the network further comprising a plurality of first signal paths each connected with a first port of a respective element, a plurality of second signal paths each connected with a second port of a respective element and at least one signal coupling means, the or each signal coupling means connecting the second signal path associated with one element with the first signal path associated with at least one other element in the network.
The invention thus provides a network comprising a plurality of elements, each element including accumulating means h

REFERENCES:
patent: 3496382 (1970-02-01), Hendrix
patent: 3691400 (1972-09-01), Askew
patent: 3950733 (1976-04-01), Cooper et al.
patent: 4518886 (1985-05-01), Clymer
patent: 4660166 (1987-04-01), Hopfield
patent: 4731747 (1988-03-01), Deuker
patent: 4782460 (1988-11-01), Spencer
International Search Report for international application No. PCT/GB87/00559.
Willwacher, G., Fahigkeiten eines assoziativen Speichersystems im Vergleich zu Gehirnfunktionen, Biol. Cybernetics, 24(4):181-198, 1976.
Fukushima, K., A Model of Associative Memory in the Brain, Kybernetik, 12:58-63, 1973.
Hirai, Y., A Learning Network Resolving Multiple Match in Associative Memory, Proceedings of Sixth International Conference on Pattern Recognition, IEEE Computer Society Press, pp. 1049-1052, 1982.
Sivilotti, M., et al., A Novel Associative Memory Implemented Using Collective Computation, 1985 Chapel Hill Conference on VLSI, H. Fuches, ed., Computer Science Press, pp. 329-342, 1986.

LandOfFree

Say what you really think

Search LandOfFree.com for the USA inventors and patents. Rate them and share your experience with other people.

Rating

Associative network and signal handling element therefor for pro does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Associative network and signal handling element therefor for pro, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Associative network and signal handling element therefor for pro will most certainly appreciate the feedback.

Rate now

     

Profile ID: LFUS-PAI-O-1042922

  Search
All data on this website is collected from public sources. Our data reflects the most accurate information available at the time of publication.