Electronic digital logic circuitry – Threshold – With field-effect transistor
Reexamination Certificate
2001-04-26
2002-12-31
Tokar, Michael (Department: 2819)
Electronic digital logic circuitry
Threshold
With field-effect transistor
C326S011000, C326S081000, C326S119000, C706S015000, C706S041000, C706S043000
Reexamination Certificate
active
06501294
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Technical Field
The present invention relates to neural networks in general, and in particular to artificial neurons utilized within a neural network. Still more particularly, the present invention relates to a neuron circuit employed as a building block of a neural network that can be implemented in an integrated circuit.
2. Description of the Prior Art
Artificial neural networks are utilized in a wide variety of applications, such as speech recognition, process control, optical character recognition, signal processing, image processing, etc. Generally speaking, an artificial neural network is made up of multiple elemental processing blocks called neuron circuits. Thus, a neuron circuit is the fundamental building block of a neural network.
Referring now to the drawings and in particular to
FIG. 1
a,
there is depicted a block diagram of a conventional neuron circuit. As shown, a neuron circuit
10
includes inputs x
1
through x
n
weighted by respective synapses w
1
through w
n
, which are accumulated together by a summing circuit
11
. The output of summing circuit
11
is then fed into a non-linear circuit
12
for generating an output
15
. In this example, the non-linear transfer function of non-linear circuit
12
is in the form of a sigmoid-shaped function that can be expressed as follows:
output=1/(1+
e
−NET
)
Many different types of non-linear transfer function can be utilized within non-linear circuit
12
. For example, the non-linear transfer function can be a binary threshold function
16
having a binary threshold, as shown in
FIG. 1
b.
In sum, a conventional neuron circuit requires a circuit for weighted input connections, a summing circuit, a multiplier circuit, and a circuit for performing a non-linear function. Because of the total size of the above-mentioned circuits, the number of conventional neuron circuits that can be manufactured on a semiconductor device is severely limited, not to mention the response time is relatively slow. Consequently, it would be desirable to provide an improved neuron circuit that is fast, simple, and inexpensive to implement.
SUMMARY OF THE INVENTION
In accordance with a preferred embodiment of the present invention, a neuron circuit that can be served as a building block for a neural network implemented in an integrated circuit includes a synapse circuit block and a neuron body circuit block. The synapse circuit block has three transistors, and the body of one of the three transistors is controlled by a weighted input. The neuron body circuit block includes a current mirror circuit, a summing circuit, and an invertor circuit. The neuron body circuit is coupled to the synapse circuit block to generate an output pulse.
All objects, features, and advantages of the present invention will become apparent in the following detailed written description.
REFERENCES:
patent: 4956564 (1990-09-01), Holler et al.
patent: 4961002 (1990-10-01), Tam et al.
patent: 5258657 (1993-11-01), Shibata et al.
patent: 5336937 (1994-08-01), Sridhar et al.
patent: 5463330 (1995-10-01), Tsuchida
patent: 5621336 (1997-04-01), Shibata et al.
patent: 5704014 (1997-12-01), Marotta et al.
patent: 5806054 (1998-09-01), Bergemont et al.
patent: 5864255 (1999-01-01), Kwok et al.
patent: 6032140 (2000-02-01), Fabbrizio et al.
patent: 6151594 (2000-11-01), Wang
patent: 6269352 (2001-07-01), Fabbrizio et al.
M.D. Voysey, and T. J. Kazmierski; “A Small, Efficient Analogue CMOS Neuron for VLSI Neural Networks,” Department of Electronics and Computer Science, Southampton University, England.
Bernstein Kerry
Rohrer Norman Jay
Bracewell & Patterson L.L.P.
Henkler Richard A.
Tan Vibol
LandOfFree
Neuron 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 Neuron circuit, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Neuron circuit will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2967474