Programmable and expandable hamming neural network circuit

Details Programmable and expandable hamming neural network circuit Programmable and expandable hamming neural network circuit

1999-04-27

2002-01-22

Davis, George B. (Department: 2122)

Data processing: artificial intelligence

Neural network

Structure

C706S020000, C706S038000

Reexamination Certificate

active

06341275

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a Hamming neural network circuit, more particularly, to a programmable and expandable Hamming neural network circuit.
2. Description of Related Art
Currently, an artificial neural network has become a fashionable and popular research topic. The neural network theory is thought to be the technique with the highest potential for solving many artificial intelligence problems. However, because a software neural network cannot satisfy many applications requiring parallel and real-time processing, the design of a very large integrated circuit (VLSI) neural network has become an important research objective.
Referring to
FIG. 1
, a typical Hamming neural network with a two-layer structure is shown. In a pattern recognition application, the first layer is used to find the matching rates between a to-be-recognized pattern and N standard patterns wherein the boxes on top of the layer are input pixels (
11
). There are N neurons (
12
) in the first layer, and each stores a standard pattern. The second layer is a winner-take-all (WTA) network (
13
) having N inputs and N outputs respectively corresponding to the N inputs. The WTA network (
13
) is used to find a standard pattern having the maximum matching rate and present a “1” on the corresponding output while the other outputs are “0”s. Although this typical Hamming neural network is easily constructed, it can only find the standard pattern which is the best match with the to-be-recognized pattern. However, with the rise in system complexity, the increase in the number of standard patterns and especially the development of expanded systems with multiple stages, the above Hamming neural network appears to be unsatisfactory. To enhance system performance, it is necessary to find two or more of the best matched standard patterns for the to-be-recognized pattern. Therefore, a novel Hamming neural network is set forth hereinafter, in which the standard patterns can be output sequentially according the degree to which they match the to-be-recognized pattern. Accordingly, m of the most matched standard patterns can be found sequentially where 1≦m≦N. This will greatly improve the system performance by increasing the recognition rate and reprocessing and reusing data in a multi-stage expanded system.
SUMMARY OF THE INVENTION
One object of the present invention is to provide a Hamming neural network circuit which can be programmed to satisfy the requirements in different applications.
Another object of the present invention is to provide a Hamming neural network circuit which can be expanded to enhance its adaptability in various applications.
In accordance with one aspect of the present invention, the Hamming neural network circuit comprises an I/O circuit for inputting and outputting a plurality of standard patterns, a bi-directional transmission gate array connected to the I/O circuit and controlled by a programming signal for transmitting the standard patterns, a plurality of standard pattern memory units connected to the bi-directional transmission gate array for storing the plurality of standard patterns, an address decoder connected to the plurality of standard pattern memory units for addressing one of the plurality of standard pattern memory units, a plurality of pattern matching calculation circuit units respectively connected to the plurality of standard pattern memory units for generating a plurality of matching rates between a to-be-recognized pattern and the plurality of standard patterns, and an expandable matching rate comparing circuit for comparing and sorting the plurality of matching rates.
Other objects, advantages, and novel features of the invention will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings.


REFERENCES:
patent: 4999525 (1991-03-01), Park et al.
patent: 5050095 (1991-09-01), Samad
patent: 5216750 (1993-06-01), Smith
patent: 5406508 (1995-04-01), Hayashibara
patent: 5517597 (1996-05-01), Aparicio, IV et al.
patent: 5537512 (1996-07-01), Hsia et al.
patent: 5630021 (1997-05-01), Li et al.
patent: 5720004 (1998-02-01), Li et al.
patent: 5999643 (1999-12-01), Shi et al.
patent: 6185331 (2001-02-01), Shi et al.
patent: 6205438 (2001-03-01), Shi et al.
patent: 6272476 (2001-08-01), Shi et al.
Lan et al.; “Analog CMOS Current-Mode Implementation of the Feedforward Neural Network with On-Chip Learning and Storage”. Proceedings of the IEEE International Conference on Neural Networks, Nov. 1995, vol. 1, p. 645-650.*
Carlos et al.; “An Extended Hamming Neural Network for Non Binary Pattern Recognition”. Proceeding of the 37thMidwest Symposium on Circuits and Systems, Aug. 1994, vol. 1, p. 607-609.*
Lu et al.; “A Modiefied Hamming Neural Network”. 1995 4thInternational Conference on Solid-State and Integrated Circuit Technology, Oct. 1995, p. 694-696.*
Nino-de-Rivera et al.; “VLSI Implementation of an Extended Hamming Neural Network for Non Binary Pattern Recognition”. Proceedings of the 38thMidwest Symposium on Circuits and Systems, Aug. 1995, vol. 2, p. 973-977.*
Huang et al.; “Neural Implementation of Fuzzy K-NN Classification for Seismic Pattern Recognition”. IEEE International Conference on Neural Networks, Jun. 1996, vol. 3, p. 1588-1593.*
Guoxing et al.; “A Modified Current Mode Hamming Neural Network for Totally Unconstrained Handwritten Numeral Recognition”. The 1998 International Joint Conference on Neural networks, May 1998, vol. 3, p. 1857-1860.*
Schmid et al.; “Hardware Realization of a Hamming Neural Network with On-Chip Learning”. Proceeding of the IEEE International Symposium on Circuits and Systems, May 1998, vol. 3, p. 191-194.

Affiliated with

Also associated with

Rating

Programmable and expandable hamming neural network circuit does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Programmable and expandable hamming neural network circuit, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Programmable and expandable hamming neural network circuit will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2857164