US5195169A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Control device for controlling learning of a neural network
Assignee: Sharp Corp
•
Filing: 1992-01-30
•
Granted: 1993-03-16
•
Expires: 2012-01-30
Abstract
Abstract A control device for controlling the learning of a neural netowrk includes a monitor for monitoring weight values of synapse connections between units of the neural netowrk during learning of the neural network so as to update these weight values. When one of the weight values satisfies a preset condition, the weight value is updated to a predetermined value such that configuration of the neural network is determined in an optimum manner.
Inventors
- 👤 Shin Kamiya
- 👤 Fumio Togawa
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-06 09:50
CPC Classifications
G; G06; G06N; G06N20/00; G06N3/00; G06N3/02; G06N3/04; G06N3/0495; G06N3/0499; G06N3/08; G06N3/082; G06N3/09
Claims Analysis
Total Claims: 8
Claim 1
Independent Claim
212 words
A system for supporting a learning process, comprising: a neural network having a plurality of layers, each layer having a plurality of units, units of one layer being connected to units of another layer by synapse connections, each synapse connection having a weight value associated therewith; and a control device, connected tos aid neural network, for controlling a learning function of said neural network; said control device including, weight input means for receiving said weight values of said synapse connections; arithmetic means, operatively connected to said weight input means, for calculating an average value for each unit of a layer, said average value being an average of said weight values of synapse connections associated with one side of said unit, decision means, operatively connected to said arithmetic means, for determining if an average value of an unit calculated by said arithmetic means is smaller than a first threshold value, for determining if a difference between target data and output data from said neural network is smaller than a second threshold value, and for outputting a signal when both determinations are affirmative, and weight update means, in response to said signal, for setting the weight values associated with the unit having an average weight value less than said first threshold value to zero.
Claim 2
Independent Claim
212 words
A system for supporting a learning process, comprising: a neural network having a plurality of layers, each layer having a plurality of units, units of one layer being connected to units of another layer by synapse connections, each synapse connection having a weight value associated therewith; and a control device, connected tos aid neural network, for controlling a learning function of said neural network; said control device including, weight input means for receiving said weight values of said synapse connections; arithmetic means, operatively connected to said weight input means, for calculating an average value for each unit of a layer, said average value being an average of said weight values of synapse connections associated with one side of said unit, decision means, operatively connected to said arithmetic means, for determining if an average value of an unit calculated by said arithmetic means is smaller than a first threshold value, for determining if a difference between target data and output data from said neural network is smaller than a second threshold value, and for outputting a signal when both determinations are affirmative, and weight update means, in response to said signal, for setting the weight values associated with the unit having an average weight value less than said first threshold value to zero.
Claim 3
Dependent Claim
42 words
The system as claimed in claim 1, wherein said control device further includes: error computing means for receiving said output data from said neural network and said target data and for computing said difference between said output data and said target data.
Claim 4
Independent Claim
227 words
A system for supporting a learning process, comprising: a neural network having a plurality of layers, each layer having a plurality of units, units of one layer being connected to units of another layer by synapse connections, each synapse connection having a weight value associated therewith; and a control device, connected tos aid neural network, for controlling a learning function of said neural network; said control device including, weight input means for receiving said weight values of said synapse connections, arithmetic means, operatively connected to said weight input means, for calculating an average value for each unit of a layer, said average value being an average of said weight values of synapse connections associated with one side of said unit, decision means, operatively connected to said arithmetic means, for determining if a first difference between an average value of one unit calculated by said arithmetic means and an average value of another unit is smaller then a first threshold value, for determining if a second difference between a target data and an output data from said neural network is smaller then a second threshold value, and for outputting a first signal when both determinations are affirmative, and weight update means, in response to said first signal, for maintaining the weight values of said one unit and for setting the weight values of said another unit to zero.
Claim 5
Independent Claim
227 words
A system for supporting a learning process, comprising: a neural network having a plurality of layers, each layer having a plurality of units, units of one layer being connected to units of another layer by synapse connections, each synapse connection having a weight value associated therewith; and a control device, connected tos aid neural network, for controlling a learning function of said neural network; said control device including, weight input means for receiving said weight values of said synapse connections, arithmetic means, operatively connected to said weight input means, for calculating an average value for each unit of a layer, said average value being an average of said weight values of synapse connections associated with one side of said unit, decision means, operatively connected to said arithmetic means, for determining if a first difference between an average value of one unit calculated by said arithmetic means and an average value of another unit is smaller then a first threshold value, for determining if a second difference between a target data and an output data from said neural network is smaller then a second threshold value, and for outputting a first signal when both determinations are affirmative, and weight update means, in response to said first signal, for maintaining the weight values of said one unit and for setting the weight values of said another unit to zero.
Claim 6
Dependent Claim
61 words
The system as claimed in claim 3 wherein said decision means outputs a second signal when said first difference is smaller than said first threshold value and said second difference is greater than said second threshold value; and said weight updating means setting the weight values of said one unit and said another unit to values based on a predetermined rule.
Claim 7
Dependent Claim
42 words
The system as claimed in claim 4, wherein said control device further includes: error computing means for receiving said output data from said neural network and said target data for computing said second difference between said output data and said target data.
Claim 8
Dependent Claim
43 words
The system as claimed in claim 3, wherein said control device further includes: error computing means for receiving said output data from said neural network and said target data and for computing said second difference between said output data and said target data.
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) | 1996-03-15 | 1996-09-14 | 1997-03-16 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2000-03-15 | 2000-09-14 | 2001-03-16 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2004-03-15 | 2004-09-14 | 2005-03-16 | $7700.00 | Window Closed / Lapsed |