Patent US5093899A successfully refreshed from Google Patents.
US5093899A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Neural network with normalized learning constant for high-speed stable learning
Assignee: Sony Corp
•
Filing: 1989-09-13
•
Granted: 1992-03-03
•
Expires: 2009-09-13
Abstract
Abstract The present invention is concerned with a signal processing system having a learning function pursuant to the back-propagation learning rule by the neural network, in which the learning rate is dynamically changed as a function of input values to effect high-speed stable learning. The signal processing system of the present invention is so arranged that, by executing signal processing for the input signals by the recurrent network formed by units each corresponding to a neuron, the features of the sequential time series pattern such as voice signals fluctuating on the time axis can be extracted through learning the coupling state of the recurrent network. The present invention modifies the prior art weight change algorithm ΔW ji (n+1) =η.δ pi .+α.W ji (n) into ΔW ji (n+1) =η.β(α pj O pi )+α.W ji (n) where β j =1/(Σ i O pi 2 +1) is used to normalize the learning constant.
Inventors
- 👤 Atsunobu Hiraiwa
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-31 11:50
CPC Classifications
G; G06; G06N; G06N3/00; G06N3/02; G06N3/08; G06N3/082; G06N3/084
Claims Analysis
Total Claims: 3
Claim 1
Independent Claim
237 words
A signal processing system comprising a signal processing section including at least an input layer, an intermediate layer and an output layer, each of said layers being made up of a plurality of signal processing units; and a learning processing section for repeatedly and sequentially computing, from said output layer towards said input layer, a coefficient W ji of coupling strength between each of said signal processing units on the basis of error data δ ji between an output value of said output layer for input signal patterns entered into said input layer and a predetermined value denoted as a teacher signal, thereby performing learning processing of said coefficient W ji of coupling strength, wherein said learning processing section includes means for computing a variant ΔW ji of said coefficient W ji of coupling strength on the basis of η·β(δ pj ·O pi ), where η stands for a learning constant, where O pi stands for values of inputs in each of said signal processing units, and where β stands for a learning variable, and for supplying said coefficient W ji of coupling strength W.sub.ji =W.sub.ji +ΔW.sub.ji to each of said signal processing units of said signal processing section, said learning processing section performing said learning processing of said coefficient W ji of coupling strength at a learning rate in which said learning constant η is normalized by said learning variable according to: β=1/O pi 2 +1).
Claim 2
Dependent Claim
237 words
A signal processing system comprising a signal processing section including at least an input layer, an intermediate layer and an output layer, each of said layers being made up of a plurality of signal processing units; and a learning processing section for repeatedly and sequentially computing, from said output layer towards said input layer, a coefficient W ji of coupling strength between each of said signal processing units on the basis of error data δ ji between an output value of said output layer for input signal patterns entered into said input layer and a predetermined value denoted as a teacher signal, thereby performing learning processing of said coefficient W ji of coupling strength, wherein said learning processing section includes means for computing a variant ΔW ji of said coefficient W ji of coupling strength on the basis of η·β(δ pj ·O pi ), where η stands for a learning constant, where O pi stands for values of inputs in each of said signal processing units, and where β stands for a learning variable, and for supplying said coefficient W ji of coupling strength W.sub.ji =W.sub.ji +ΔW.sub.ji to each of said signal processing units of said signal processing section, said learning processing section performing said learning processing of said coefficient W ji of coupling strength at a learning rate in which said learning constant η is normalized by said learning variable according to: β=1/O pi 2 +1).
Claim 3
Dependent Claim
68 words
The signal processing system according to claim 1 wherein delay means are provided in each of said signal processing units of said intermediate and output layers of said signal processing section and wherein said signal processing section is arranged as a recurrent network including loop and feedback circuitry connecting an output of said signal processing section through said delay means to an input of said signal processing section.
References Cited (Backward Citations)
No citation records indexed.
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1995-03-03 | 1995-09-02 | 1996-03-03 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 1999-03-03 | 1999-09-02 | 2000-03-03 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2003-03-03 | 2003-09-02 | 2004-03-03 | $7700.00 | Window Closed / Lapsed |