home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai.neural-nets
- Path: sparky!uunet!world!srctran
- From: srctran@world.std.com (Gregory Aharonian)
- Subject: Abstracts for new neural network patents
- Message-ID: <BzGn1A.C32@world.std.com>
- Organization: The World Public Access UNIX, Brookline, MA
- Date: Fri, 18 Dec 1992 15:00:45 GMT
- Lines: 317
-
-
- The following is my periodic posting of the abstracts and stat data
- for new neural network patents. The full text and diagrams of any of these
- patents can be ordered from the patent office for $3 in printed form. Also,
- for $20, I will provide a machine readable version of the abstract, text,
- claims and references for any neural network patent.
-
- Greg Aharonian
- Source Translation & Optimization
- srctran@world.std.com
- 617-489-3727
- =============================================================================
- For a very good article on the legal issues involving neural network
- patents, see the article "Intellectual Property Protection for Neural
- Networks" by Donald Wenskay, Neural Networks, 3, pp. 229-236, 1990.
- =============================================================================
- For those interested in applying for a neural network patent, I freely
- provide a LaTeX template for a patent application. Please email a request,
- and I'll send it to you. Contains most of the boilerplate you need.
- =============================================================================
- To date, I have yet to see any articles in the trade journals on either
- companies signing cross-licensing agreements to share their neural network
- patents, or patent infringement lawsuits dealing with neural network patents.
- To me, this means that no one is making much money with neural networks (other
- than selling software), and that the financial return on acquiring a patent
- is less than the cost of applying for and maintaining a patent.
- If anyone hears of either occurrence, please let me know.
- =============================================================================
-
- ==============================================================================
- 5,166,539 [IMAGE AVAILABLE] Nov. 24, 1992
-
- Neural network circuit
-
- INVENTOR: Kuniharu Uchimura, Kanagawa, Japan
- Osamu Saito, Kanagawa, Japan
- Yoshihito Amemiya, Tokyo, Japan
- Atsushi Iwata, Tokyo, Japan
- ASSIGNEE: Nippon Telegraph and Telephone Corporation, Tokyo, Japan
- APPL-NO: 07/727,065
- DATE FILED: Jul. 8, 1991
- INT-CL: [5] H03K 19/08
- US-CL-ISSUED: 307/201, 464
- US-CL-CURRENT: 307/201, 464
- SEARCH-FLD: 307/201, 464, 465; 364/513, 807; 395/21-24
- ART-UNIT: 252
- PRIM-EXMR: Eugene R. LaRoche
- ASST-EXMR: A. Zarabian
- LEGAL-REP: Finnegan, Henderson Farabow, Garrett and Dunner
-
- ABSTRACT:
- A neural network circuit, in which a number n of weight coefficients (Wl-wn)
- corresponding to a number n of inputs are provided, subtraction circuits
- determine the difference between inputs and the weight coefficients in each
- input terminal, the result thereof is inputted into absolute value circuits,
- all calculation results of the absolute value circuts corresponding to the
- inputs and the weight coefficients are inputted into an addition circuit and
- accumulated, and this accumulation result determines the output value. The
- threshold value circuit, which determines the final output value, has
- characteristics of a step function pattern, a polygonal line pattern, or a
- sigmoid function pattern, depending on the object. In the case in which a
- neural network circuit is realized by means of digital circuits, the absolute
- value circuits can comprise simply EX-OR logic (exclusive OR) gates.
- Furthermore, in the case in which the input terminals have two input paths
- and two weight coefficients corresponding to each input path, the neuron
- circuits form a recognition area having a flexible shape which is controlled
- by the weight coefficients. Neuron circuits are widely used in pattern
- recognition; neuron circuits react to a pattern inputted into the input layer
- and recognition is thereby conducted.
- 5 Claims, 40 Drawing Figures
- ==============================================================================
- 5,166,896 [IMAGE AVAILABLE] Nov. 24, 1992
-
- Discrete cosine transform chip using neural network concepts for calculating
- values of a discrete cosine transform function
-
- INVENTOR: Ho-Sun Jeong, Taegu, Republic of Korea
- Je-kwang Ryu, Taegu, Republic of Korea
- ASSIGNEE: Samsung Electronics Co., Ltd., Kyunggi-do, Republic of Korea
- APPL-NO: 07/659,089
- DATE FILED: Feb. 22, 1991
- FRN-PRIOR: Republic of Korea 90-4513 Apr. 3, 1990
- INT-CL: [5] G06F 7/38
- US-CL-ISSUED: 364/725
- US-CL-CURRENT: 364/725
- SEARCH-FLD: 364/725, 748, 768, 754; 395/21
- ART-UNIT: 236
- PRIM-EXMR: Long T. Nguyen
- LEGAL-REP: Cushman, Darby & Cushman
-
- ABSTRACT:
- A discrete cosine transform chip includes circuits using neural network
- concepts that have parallel processing capability as well as conventional
- digital logic circuits. In particular, the discrete cosine transform chip
- includes a cosine term processing portion, a multiplier, an adder, a
- subtractor, and two groups of latches. The multiplier, the adder and the
- subtractor incorporated in the discrete cosine transform chip use
- unidirectional feed back neural network models.
- 12 Claims, 10 Drawing Figures
- ==============================================================================
- 5,166,927 [IMAGE AVAILABLE] Nov. 24, 1992
-
- Adaptive pathfinding neutral network for a packet communication system
-
- INVENTOR: Ichiro Iida, Yokohama, Japan
- Akira Chugo, Tokyo, Japan
- ASSIGNEE: Fujitsu Limited, Kanagawa, Japan
- APPL-NO: 07/455,323
- DATE FILED: Mar. 6, 1990
- PCT-FILED: Jul. 6, 1989
- PCT-NO: PCT/JP89/00684
- 371-DATE: Mar. 6, 1990
- 102(E)-DATE: Mar. 6, 1990
- PCT-PUB-NO: WO90/00842
- PCT-PUB-DATE: Jan. 25, 1990
- INT-CL: [5] H04Q 11/04
- US-CL-ISSUED: 370/60, 94.1, 94.3, 54; 395/21
- US-CL-CURRENT: 370/60, 54, 94.1, 94.3; 395/21
- SEARCH-FLD: 370/94.1, 95.3, 16, 60, 94.3, 54, 79; 395/22, 21
- ART-UNIT: 263
- PRIM-EXMR: Douglas W. Olms
- ASST-EXMR: Shick Hom
- LEGAL-REP: Staas & Halsey
-
- ABSTRACT:
-
- An adaptive routing system is used in a network for performing a
- communication in a packet form obtained by adding address data and data
- length to the transmission information. This connects nodes having input
- ports and output ports arranged in a distributed manner, by the input link
- and output link. This system enables the network to assign the neuron
- elements to input ports and output ports of respective nodes one by one,
- inputs a network state to the neuron element to evaluate the traffic
- condition of the network and determines the output port assigned to the
- neuron element having the "1" output from among the neuron elements assigned
- to the output ports of the respective nodes when the outputs of respective
- neuron elements are not changed after reaching the balanced state, thereby
- enabling the output of said output port to be produced in the optimum output
- direction of the packet from respective nodes.
- 64 Claims, 39 Drawing Figures
- ==============================================================================
- 5,166,938 [IMAGE AVAILABLE] Nov. 24, 1992
-
- Error correction circuit using a design based on a neural network model
- comprising an encoder portion and a decoder portion
-
- INVENTOR: Ho-Sun Chung, Taegu, Republic of Korea
- ASSIGNEE: Samsung Electronics Co., Ltd., Kyunggi, Republic of Korea
- APPL-NO: 07/549,931
- DATE FILED: Jul. 9, 1990
- FRN-PRIOR: Republic of Korea 90-9552 Jun. 27, 1990
- INT-CL: [5] G06F 11/08
- US-CL-ISSUED: 371/37.1; 395/27
- US-CL-CURRENT: 371/37.1; 395/27
- SEARCH-FLD: 395/27, 24, 22; 371/37.1
- ART-UNIT: 233
- PRIM-EXMR: Robert W. Beausoliel
- ASST-EXMR: Henry C. Lebowitz
- LEGAL-REP: Cushman, Darby & Cushman
-
- ABSTRACT:
- An error correction circuit is provided which uses NMOS and PMOS synapses to
- form network type responses to a coded multi-bit input. Use of MOS technology
- logic in error correction circuits allows such devices to be easily
- interfaced with other like technology circuits without the need to use
- distinct interface logic as with conventional error correction circuitry.
- 4 Claims, 22 Drawing Figures
- ==============================================================================
- 5,167,006 [IMAGE AVAILABLE] Nov. 24, 1992
-
- Neuron unit, neural network and signal processing method
-
- INVENTOR: Toshiyuki Furuta, Yokohama, Japan
- Hiroyuki Horiguchi, Yokohama, Japan
- Hirotoshi Eguchi, Yokohama, Japan
- Yutaka Ebi, Yokohama, Japan
- Tatsuya Furukawa, Yokohama, Japan
- Yoshio Watanabe, Kawasaki, Japan
- Toshihiro Tsukagoshi, Itami, Japan
- ASSIGNEE: Ricoh Company, Ltd., Japan
- APPL-NO: 07/629,632
- DATE FILED: Dec. 18, 1990
- INT-CL: [5] G06F 15/18
- US-CL-ISSUED: 395/11, 27
- US-CL-CURRENT: 395/11, 27
- SEARCH-FLD: 395/24, 26, 27, 11
- ART-UNIT: 238
- PRIM-EXMR: Allen R. MacDonald
- LEGAL-REP: Mason, Fenwick & Lawrence
-
- ABSTRACT:
- A neuron unit processes a plurality of input signals and outputs an output
- signal which is indicative of a result of the processing. The neuron unit
- includes input lines for receiving the input signals, a forward process part
- including a supplying part for supplying weight functions and an operation
- part for carrying out an operation on each of the input signals using one of
- the weight functions and for outputting the output signal, and a
- self-learning part including a generating part for generating new weight
- functions based on errors between the output signal of the forward process
- part and teaching signals and a varying part for varying the weight functions
- supplied by the supplying part of the forward process part to the new weight
- functions generated by the generating part.
- 72 Claims, 76 Drawing Figures
- ==============================================================================
- 5,167,007 [IMAGE AVAILABLE] Nov. 24, 1992
-
- Multilayered optical neural network system
-
- INVENTOR: Haruyoshi Toyoda, Fukuroi, Japan
- ASSIGNEE: Hamamatsu Photonics K.K., Japan
- APPL-NO: 07/598,173
- DATE FILED: Oct. 12, 1990
- FRN-PRIOR: Japan 1-269718 Oct. 17, 1989
- INT-CL: [5] G06F 15/18
- US-CL-ISSUED: 395/25
- US-CL-CURRENT: 395/25
- SEARCH-FLD: 364/513; 395/25
- ART-UNIT: 238
- PRIM-EXMR: Allen R. MacDonald
- LEGAL-REP: Oliff & Berridge
-
- ABSTRACT:
- A multilayered optical neural network system comprise an input layer, an
- output layer, at least one hidden layer provided between the input layer and
- the output layer, a memory matrix holding device provided between the
- respective layers for holding weighted couplings between the layers, a
- correlation operating device for optically computing a correlation between an
- output optical pattern from the previous layer and the memory matrix pattern,
- an output function operating device for implementing optical computing of an
- output function corresponding to a result of the correlation operation, and a
- memory matrix correcting device provided between the respective layers for
- optically correcting a memory matrix held in the memory matrix holding device
- by a learning operation, whereby the system is capable of two-dimensional
- optical computing of all data transfers and operations and executing a great
- amount of computing without use of holograms.
- 9 Claims, 26 Drawing Figures
- ==============================================================================
- 5,167,008 [IMAGE AVAILABLE] Nov. 24, 1992
-
- Digital circuitry for approximating sigmoidal response in a neural network
- layer
-
- INVENTOR: William E. Engeler, Scotia, NY
- ASSIGNEE: General Electric Company, Schenectady, NY
- APPL-NO: 07/752,290
- DATE FILED: Aug. 29, 1991
- REL-US-DATA: Division of Ser. No. 628,257, Dec. 14, 1990, Pat. No.
- 5,115,492, May 19, 1992.
- INT-CL: [5] G06F 15/18
- US-CL-ISSUED: 395/27; 364/715.01, 602; 395/24
- US-CL-CURRENT: 395/27; 364/602, 715.01; 395/24
- SEARCH-FLD: 395/27, 24; 364/715.01, 716, 729, 602
- ART-UNIT: 238
- PRIM-EXMR: Michael R. Fleming
- ASST-EXMR: Robert W. Downs
- LEGAL-REP: Marvin Snyder
-
- ABSTRACT:
- A plurality of neural circuits are connected in a neural network layer for
- generating their respective digital axonal responses to the same plurality of
- synapse input signals. Each neural circuit includes digital circuitry for
- approximating a sigmoidal response connected after respective circuitry for
- performing a weighted summation of the synapse input signals to generate a
- weighted summation result in digital form. In this digital circuitry the
- absolute value of the digital weighted summation result is first determined.
- Then, a window comparator determines into which of a plurality of amplitude
- ranges the absolute value of the weighted summation result falls. A digital
- intercept value and a digital slope value are selected in accordance with the
- range into which the absolute value of the weighted summation result falls.
- The absolute value of the digital weighted summation result is multiplied by
- the selected digital slope value to generate a digital product; and the
- digital intercept value is added to the digital product to generate an
- absolute value representation of a digital axonal response. The polarity of
- the weighted summation result is determined, and the same polarity is
- assigned to the absolute value representation of the digital axonal response,
- thereby to generate the digital axonal response.
- 4 Claims, 38 Drawing Figures
- ==============================================================================
- 5,167,009 [IMAGE AVAILABLE] Nov. 24, 1992
-
- On-line process control neural network using data pointers
-
- INVENTOR: Richard D. Skeirik, Newark, DE
- ASSIGNEE: E. I. Du Pont de Nemours & Co. (Inc.), Wilmington, DE
- APPL-NO: 07/562,388
- DATE FILED: Aug. 3, 1990
- INT-CL: [5] G06F 15/18
- US-CL-ISSUED: 395/27, 22, 68, 11, 906
- US-CL-CURRENT: 395/27, 11, 22, 68, 906
- SEARCH-FLD: 364/513, 148, 149, 150, 151, 164, 165, 807, 200, 474.15, 500,
- 501, 502, 503; 395/22, 26, 27, 906, 907, 914, 915, 11, 68
- ART-UNIT: 238
- PRIM-EXMR: Allen R. MacDonald
- LEGAL-REP: Sterne, Kessler, Goldstein & Fox
-
- ABSTRACT:
- An on-line process control neural network using data pointers allows the
- neural network to be easily configured to use data in a process control
- environment. The inputs, outputs, training inputs and errors can be retrieved
- and/or stored from any available data source without programming. The user of
- the neural network specifies data pointers indicating the particular computer
- system in which the data resides or will be stored; the type of data to be
- retrieved and/or stored; and the specific data value or storage location to
- be used. The data pointers include maximum, minimum, and maximum change
- limits, which can also serve as scaling limits for the neural network. Data
- pointers indicating time-dependent data, such as time averages, also include
- time boundary specifiers. The data pointers are entered by the user of the
- neural network using pop-up menus and by completing fields in a template. An
- historical database provides both a source of input data and a storage
- function for output and error data.
- 24 Claims, 34 Drawing Figures
-
- --
- **************************************************************************
- Greg Aharonian
- Source Translation & Optimiztion
- P.O. Box 404, Belmont, MA 02178
-