US5909676A
Expired / Lapsed
US
⏱️ 0.0 years remaining
System for controlling an object and medium using neural networks
Assignee: Toshiba Corp
•
Filing: 1997-02-27
•
Granted: 1999-06-01
•
Expires: 2017-02-27
Abstract
Abstract A function for compensating an error between a teacher signal and an output signal with a weight value is defined. A multilayered neural network is changed so that the function becomes minimum. Thus, the multilayered neural network can be adaptively controlled.
Inventors
- 👤 Makoto Kano
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-04 00:58
CPC Classifications
G; G06; G06N; G06N3/00; G06N3/02; G06N3/08; G06N3/04; G06N3/0499; G06N3/09
Claims Analysis
Total Claims: 42
Claim 1
Independent Claim
142 words
A controlling apparatus comprising: a multilayered neural network designating a range of a permissible error between a teacher signal and an output signal; means for calculating the error between the teacher signal and the output signal received from said multilayered neural network; weight designating means for designating a weight corresponding to the range; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein the function is defined as follows: ##EQU15## where t k is a teacher signal; o k is an output signal; s k and s 1 are weights; k and l are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
Claim 2
Dependent Claim
142 words
A controlling apparatus comprising: a multilayered neural network designating a range of a permissible error between a teacher signal and an output signal; means for calculating the error between the teacher signal and the output signal received from said multilayered neural network; weight designating means for designating a weight corresponding to the range; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein the function is defined as follows: ##EQU15## where t k is a teacher signal; o k is an output signal; s k and s 1 are weights; k and l are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
Claim 3
Dependent Claim
111 words
A controlling apparatus, comprising: a multilayered neural network designating range of a permissible error between a teacher signal and an output signal; means for calculating the error between the teacher signal and the output signal received from said multilayered neural network; weight designating means for designating a weight corresponding to the range; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein said weight designating means designates a value proportional to the inverse number of the designated range as a weight.
Claim 4
Dependent Claim
111 words
A controlling apparatus, comprising: a multilayered neural network designating range of a permissible error between a teacher signal and an output signal; means for calculating the error between the teacher signal and the output signal received from said multilayered neural network; weight designating means for designating a weight corresponding to the range; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein said weight designating means designates a value proportional to the inverse number of the designated range as a weight.
Claim 5
Dependent Claim
99 words
A controlling apparatus, comprising: a multilayered neural network; means for calculating an error between a teacher signal and an output signal received from said multilayered neural network; means for designating a weight corresponding to the error; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein said weight designating means designates a value proportional to the inverse number of a square value of the calculated error as a weight.
Claim 6
Dependent Claim
99 words
A controlling apparatus, comprising: a multilayered neural network; means for calculating an error between a teacher signal and an output signal received from said multilayered neural network; means for designating a weight corresponding to the error; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein said weight designating means designates a value proportional to the inverse number of a square value of the calculated error as a weight.
Claim 7
Dependent Claim
99 words
A controlling apparatus, comprising: a multilayered neural network; means for calculating an error between a teacher signal and an output signal received from said multilayered neural network; means for designating a weight corresponding to the error; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein said weight designating means designates a value proportional to the inverse number of the absolute value of the calculated error as a weight.
Claim 8
Dependent Claim
99 words
A controlling apparatus, comprising: a multilayered neural network; means for calculating an error between a teacher signal and an output signal received from said multilayered neural network; means for designating a weight corresponding to the error; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein said weight designating means designates a value proportional to the inverse number of the absolute value of the calculated error as a weight.
Claim 9
Dependent Claim
95 words
A controlling apparatus, comprising: a multilayered neural network; means for calculating an error between a teacher signal and an output signal received from said multilayered neural network; means for designating a weight corresponding to the error; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein said weight designating means designates a value proportional to a square value of the calculated error as a weight.
Claim 10
Dependent Claim
95 words
A controlling apparatus, comprising: a multilayered neural network; means for calculating an error between a teacher signal and an output signal received from said multilayered neural network; means for designating a weight corresponding to the error; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein said weight designating means designates a value proportional to a square value of the calculated error as a weight.
Claim 11
Dependent Claim
95 words
A controlling apparatus, comprising: a multilayered neural network; means for calculating an error between a teacher signal and an output signal received from said multilayered neural network; means for designating a weight corresponding to the error; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein said weight designating means designates a value proportional to the absolute value of the calculated error as a weight.
Claim 12
Dependent Claim
95 words
A controlling apparatus, comprising: a multilayered neural network; means for calculating an error between a teacher signal and an output signal received from said multilayered neural network; means for designating a weight corresponding to the error; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function becomes minimum, wherein said weight designating means designates a value proportional to the absolute value of the calculated error as a weight.
Claim 13
Dependent Claim
134 words
A controlling method, comprising the steps of: designating a range of a permissible error between a teacher signal and an output signal received from a multilayered neural network; calculating an error between the teacher signal and the output signal; designating a weight corresponding to the designated range; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the function is defined as follows: ##EQU16## where t k is a teacher signal; o k is an output signal; s k and s 1 are weights; k and l are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
Claim 14
Dependent Claim
134 words
A controlling method, comprising the steps of: designating a range of a permissible error between a teacher signal and an output signal received from a multilayered neural network; calculating an error between the teacher signal and the output signal; designating a weight corresponding to the designated range; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the function is defined as follows: ##EQU16## where t k is a teacher signal; o k is an output signal; s k and s 1 are weights; k and l are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
Claim 15
Dependent Claim
101 words
A controlling method, comprising the steps of: designating a range of a permissible error between a teacher signal and an output signal received from a multilayered neural network; calculating an error between the teacher signal and the output signal; designating a weight corresponding to the designated range; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the inverse number of the designated range.
Claim 16
Dependent Claim
101 words
A controlling method, comprising the steps of: designating a range of a permissible error between a teacher signal and an output signal received from a multilayered neural network; calculating an error between the teacher signal and the output signal; designating a weight corresponding to the designated range; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the inverse number of the designated range.
Claim 17
Dependent Claim
90 words
A controlling method of controlling a mutilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the inverse number of a square value or the calculated error.
Claim 18
Dependent Claim
90 words
A controlling method of controlling a mutilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the inverse number of a square value or the calculated error.
Claim 19
Dependent Claim
90 words
A controlling method of controlling a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the inverse number of the absolute value of the calculated error.
Claim 20
Dependent Claim
90 words
A controlling method of controlling a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the inverse number of the absolute value of the calculated error.
Claim 21
Dependent Claim
86 words
A controlling method of controlling a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to a square value of the calculated error.
Claim 22
Dependent Claim
86 words
A controlling method of controlling a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to a square value of the calculated error.
Claim 23
Dependent Claim
86 words
A controlling method of controlling a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the absolute value of the calculated error.
Claim 24
Dependent Claim
86 words
A controlling method of controlling a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the absolute value of the calculated error.
Claim 25
Independent Claim
136 words
A medium having a program, comprising the steps of: designating a range of a permissible error between a teacher signal and an output signal received from a multilayered neural network; calculating an error between the teacher signal and the output signal; designated a weight corresponding to the designated range; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the function is defined as follows: ##EQU17## where t k is a teacher signal; o k is an output signal; s k and s 1 are weights; k and l are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
Claim 26
Independent Claim
136 words
A medium having a program, comprising the steps of: designating a range of a permissible error between a teacher signal and an output signal received from a multilayered neural network; calculating an error between the teacher signal and the output signal; designated a weight corresponding to the designated range; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the function is defined as follows: ##EQU17## where t k is a teacher signal; o k is an output signal; s k and s 1 are weights; k and l are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
Claim 27
Independent Claim
102 words
A medium having a program, comprising the steps of: designating a range of a permissible error between a teacher signal and an output signal received from a multilayered neural network; calculating an error between the teacher signal and the output signal; designated a weight corresponding to the designated range; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as value proportional to the inverse number of the designated range.
Claim 28
Independent Claim
102 words
A medium having a program, comprising the steps of: designating a range of a permissible error between a teacher signal and an output signal received from a multilayered neural network; calculating an error between the teacher signal and the output signal; designated a weight corresponding to the designated range; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as value proportional to the inverse number of the designated range.
Claim 29
Independent Claim
91 words
A medium having a program of a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the inverse number of a square value of the calculated error.
Claim 30
Independent Claim
91 words
A medium having a program of a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the inverse number of a square value of the calculated error.
Claim 31
Independent Claim
91 words
A medium having a program of a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the inverse number of the absolute value of the calculated error.
Claim 32
Independent Claim
91 words
A medium having a program of a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the inverse number of the absolute value of the calculated error.
Claim 33
Independent Claim
86 words
A medium having a program of a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to a square value of the calculated error.
Claim 34
Independent Claim
86 words
A medium having a program of a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to a square value of the calculated error.
Claim 35
Independent Claim
86 words
A medium having a program of a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the absolute of the calculated error.
Claim 36
Independent Claim
86 words
A medium having a program of a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the weight is designated as a value proportional to the absolute of the calculated error.
Claim 37
Dependent Claim
125 words
A controlling apparatus, comprising: a multilayered neural network; means for calculating an error between a teacher signal and an output signal received from said multilayered neural network; means for designating a weight corresponding to the error; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function become minimum, wherein the function is defined as follows: ##EQU18## where t k is a teacher signal, o k is an output signal; s k and s 1 are weights; k and 1 are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
Claim 38
Dependent Claim
125 words
A controlling apparatus, comprising: a multilayered neural network; means for calculating an error between a teacher signal and an output signal received from said multilayered neural network; means for designating a weight corresponding to the error; and means for defining a function reflecting the value of the error and the value of the designated weight and for prosecuting the learning process of the multilayered neural network so that the value of the function become minimum, wherein the function is defined as follows: ##EQU18## where t k is a teacher signal, o k is an output signal; s k and s 1 are weights; k and 1 are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
Claim 39
Dependent Claim
119 words
A controlling method of controlling a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the function is defined as follows: ##EQU19## where t k is a teacher signal; o k is an output signal; s k and s 1 are weights; k and 1 are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
Claim 40
Dependent Claim
119 words
A controlling method of controlling a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the function is defined as follows: ##EQU19## where t k is a teacher signal; o k is an output signal; s k and s 1 are weights; k and 1 are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
Claim 41
Independent Claim
120 words
A medium having a program of a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the function is defined as follows: ##EQU20## where t k is a teacher signal; o k is an output signal; s k and s 1 are weights; k and l are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
Claim 42
Independent Claim
120 words
A medium having a program of a multilayered neural network, comprising the steps of: calculating an error between a teacher signal and an output signal; designating a weight corresponding to the calculated error; defining a function reflecting the value of the calculated error and the value of the designated weight; and prosecuting the learning process of the multilayered neural network so that the value of the defined function becomes minimum, wherein the function is defined as follows: ##EQU20## where t k is a teacher signal; o k is an output signal; s k and s 1 are weights; k and l are variable natural numbers that are 1 or larger and m is a dimensional number of the output signal.
References Cited (Backward Citations)
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 2002-05-31 | 2002-11-30 | 2003-06-01 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2006-05-31 | 2006-11-30 | 2007-06-01 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2010-05-31 | 2010-11-30 | 2011-06-01 | $7700.00 | Window Closed / Lapsed |