CLASSIFICATION ESTIMATING SYSTEM AND CLASSIFICATION ESTIMATING PROGRAM

• US 20120209134A1
• Filed: 01/13/2012
• Published: 08/16/2012
• Est. Priority Date: 07/15/2009
• Status: Active Grant

• Current Assignee: University of Tsukuba
• Original Assignee: University of Tsukuba
• Inventors: Morita, Masahiko, Kawata, Hiroshi
• Granted Patent: US 9,392,954 B2
• Time in Patent Office: Days
• US Class Current: 600/546
• CPC Class Codes: A61B 5/316 Modalities, i.e. specific diagnostic methods
• CPC Class Codes: A61B 5/389 Electromyography [EMG]
• CPC Class Codes: A61B 5/6824 Arm or wrist
• CPC Class Codes: A61B 5/7235 Details of waveform analysis A61B5/349 takes precedence
• CPC Class Codes: A61B 5/7267 involving training the classification device
• CPC Class Codes: G06F 3/015 Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection
• CPC Class Codes: G06F 3/017 Gesture based interaction, e.g. based on a set of recognized hand gestures interaction based on gestures traced on a digitiser G06F3/04883
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06V 40/28 Recognition of hand or arm movements, e.g. recognition of deaf sign language static hand signs G06V40/113
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients

Abstract

A classification estimating system can include an input element group; an intermediate element group into which are input values of first intermediate variables for which a first multiple desensitization, including an accumulation of values of each input element of a first input element group and each value of a first output sensitivity and each value of a first multiplex output sensitivity, has been carried out and calculated; an output element group into which is input a value of an output variable calculated based on a value of each intermediate element of a first intermediate element group and a value of each connection weight; a classification information estimating module for estimating classification information based on pre-stored correspondence relationship information and a value of the output variable; and a classification information display.

Claims

1. A classification estimating system, comprising:

an input element group comprising a first input element group comprising a plurality of input elements into which are input values of the first input variable based on a first information and a second input element group consisted of a plurality of input elements into which are input values of the second input variable based on a second information;

an intermediate element group provided with a first intermediate element group comprising a plurality of intermediate elements into which are input a first intermediate variables which are calculated from the value of each input element of said first input element group, wherein a first multiple desensitization including at least an accumulation of each value of a first output sensitivity and each value of a first multiplex output sensitivity has been carried out while some values become zero;

wherein each value of a first output sensitivity is calculated according to each value of input element of said first input element group and each value of input element of said second input element group in order for some values to become zero, and each value of a first multiplex output sensitivity is calculated according to each pre-set value of the element of the multiplex element group in order for some values to become zero;

an output element group comprising a plurality of output elements into which are input the values of output variables which are calculated according to each value of said intermediate element of first intermediate variables and each pre-set value of connection weights corresponding to the importance of each value of said immediate elements.

a correspondence relationship information memorizing means to memorize correspondence relationship information, which specifies the correspondence relationship between a plurality of classification information, as the object of classification, and the values of said output variables corresponding to the classification information.

a classification information estimating means to estimate said classification information corresponding to said first and second information that are input according to said correspondence relationship information pre-stored in said correspondence relationship information memorizing means and the calculated values of said output variables.

a classification information display means to display said classification information that is estimated according to said classification information estimating means.

2. A classification estimating system according to claim 1, comprising:

said intermediate element group provided with a second intermediate element group comprising a plurality of intermediate elements into which are input a second intermediate variables which are calculated from the values of each input element of said second input element group, wherein a second multiple desensitization including at least an accumulation of each value of a second output sensitivity and each value of a second multiplex output sensitivity has been carried out while some values become zero;

wherein each value of a second output sensitivity is calculated according to each value of input element of said first input element group and each value of input elements of said second input element group in order for some values to become zero, and each value of a first multiplex output sensitivity is calculated according to each pre-set value of the elements of a plurality of element group in order for some values to become zero;

said output element group into which are input the values of output variables which are calculated according to each value of the elements of said intermediate variables and each pre-set value of connection weights corresponding to the importance of each value of said immediate elements.

3. A classification estimating system according to any one of claim 1 or 2, comprising:

said classification information estimated according to said classification information estimating means;

a connection weight training means to study each value of said connection weights by way of updating each value of said connection weights based on the difference from the real classification information extracted from said input first and second information.

4. A classification estimating system according to any one of claims 1 to 3, wherein:

each value of element of said multiplex element group is calculated according to said first and second information.

5. A classification estimating system according to any one of claims 1 to 4, comprising:

measuring devices comprising a first measuring device which measures said biosignals at the first measuring section arranged on said subject; and

a second measuring device which measures said biosignals at the second measuring section arranged on said subject which is different from the first measuring section, wherein said classification information is taken as the operational decision of a subject and together with said operational decision, said first as well as the second information is taken as the biosignals of said subject.

6. A classification estimating system according to claim 5, comprising:

said measuring devices comprising said first measuring device which measures said surface EMG signals at said first measuring section arranged on said skin; and

said second measuring device which measures said surface EMG signals at said second measuring section arranged on said skin, wherein among the EMG signals indicating the information on potential change based on the muscle activity of the muscle group of said subject, together with said operational decision, said EMG signals measured at the skin of said subject are taken as said surface EMG signals, whereas said EMG signals are taken as the surface EMG signals;

an integrated value calculating method for calculating the integrated values of said surface EMG signals measured by said first measuring device as the first integrated value, as well as the integrated values of said surface EMG signals measured by said second measuring device as the second integrated values;

an average integrated value calculating method for calculating the average values of the first integrated values as the first average integrated values over a certain preset shift time, as well as the average values of the second integrated values as the second average integrated value over said shift time within a certain pre-set time frame;

said input element group comprising said first input element group into which are input the values of said first input variable over said shift time according to said first average integrated value, and said second input element group into which are input the values of said second input variable over said shift time according to said second average integrated value, wherein said first information is taken as said first average integrated value and said second information is taken as said second average integrated value;

7. A classification estimating system according to claim 5 or 6, comprising:

said classification information estimating means to estimate said classification information including, information on types to specify the information of motion types according to said operational decision, as well as information on force as the information of force of said motions.

8. A classification estimating system according to claim 7, wherein the classification estimating system comprises:

said classification information estimating means to estimate said classification information comprising said velocity information which, as velocity information, is the information on velocity of said motion, based on said force information on types to specify the information of motion types according to said operational decision, and the information on force as the information of force of said motions.

9. A classification estimating system according to claim 8, wherein the classification estimating system comprises:

said output element group comprising the first output element group consisting of a plurality of output elements into which are input the values of the first output variables corresponding to the first operational decision and the second output element group consisting of a plurality of output elements into which are input the values of the second output variables corresponding to the second operational decision among the values of said output variables;

said correspondence relationship information memorizing means to memorize said correspondence relationship information which specifies the correspondence relationship between said information on types of said first operational decision as well as said velocity information, and the value of said first output variables; and

the correspondence relationship between said information on types of said second operational decision as well as said velocity information, and the value of said second output variables;

said classification information estimating means to estimate said type information of said classification information as well as said velocity information according to said first and second information into which are input the values based on said correspondence relationship information memorized in said correspondence relationship memorizing means as well as the calculated values of said first output variables and said second output variables;

said classification information estimating means to indicate said type information of said classification information;

as well as said velocity information according to said classification information estimating means.

10. A classification estimating program, wherein a computer is functioned as:

an input value memorizing means which memorize the values of said first input variables as well as the values of said second input variables with regard to the input element group which comprises a first input element group derived from a plurality of input elements into which are input values of the first input variables based on the first information;

a second input element group derived from a plurality of input elements into which are input values of the second input variables based on the first information;

an intermediate value calculating means to calculate the values of said first intermediate variables with regard to the intermediate element group provided with a first intermediate element group which comprises a plurality of intermediate elements into which are input a first intermediate variables which are calculated from the value of each input element of said first input element group, for which a first multiple desensitization including at least an accumulation of each value of a first output sensitivity and each value of a first multiplex output sensitivity has been carried out while some values become zero;

wherein each value of a first output sensitivity is calculated according to each value of input element of said first input element group and each value of input element of said second input element group in order for some values to become zero, and each value of a first multiplex output sensitivity is calculated according to each pre-set value of the element of the multiplex element group in order for some values to become zero;

an output calculating means to calculate the values of said output variables regarding the output element group comprising a plurality of output elements into which are input the values of output variables which are calculated according to each value of said intermediate element of first intermediate variables and each pre-set value of connection weights corresponding to the importance of each value of said immediate elements;

a correspondence relationship information memorizing means to memorize correspondence relationship information which specifies the correspondence relationship between the multiple classification information as the object of classification; and

the values of said output variables corresponding to the classification information;

a classification information estimating means to estimate said classification information corresponding to said first and second information that are input based on said correspondence relationship information pre-stored in said correspondence relationship information memorizing means and the calculated values of said output variables;

a classification information display means to display said classification information estimated according to said classification information estimating means.