Classification estimating system and classification estimating program
First Claim
1. A classification estimating system for a subject'"'"'s motion, based on a multi-layered neural network comprising:
- a first measuring device configured to measure first information as a first biosignal according to a first measuring section of the subject;
a second measuring device configured to measure second information as a second biosignal according to a second measuring section of the subject;
a computer-readable memory storing executable instructions; and
a processor in communication with the computer-readable memory and the first and second measuring devices, wherein the processor is programmed by the executable instructions to at least;
use the multi-layered neural network, the multi-layered neural network comprising an input element group in an input layer of the multi-layered neural network, an intermediate element group in a hidden layer of the multi-layered neural network and an output element group in an output layer of the multi-layered neural network, for estimating an action of the subject;
wherein the input element group of the multi-layered neural network comprises a first input element group and a second input element group, the first element group comprising a first plurality of input elements, the second element group comprising a second plurality of input elements;
determine values of a first input variable based on the first information;
provide each value of the first input variable to each of the first plurality of input elements of the first input element group;
determine values of a second input variable based on the second information;
provide each value of the second input variable to each of the second plurality of input elements of the second input element group;
wherein the intermediate element group of the multi-layered neural network comprises a first intermediate element group, the first intermediate element group comprising a plurality of intermediate elements;
set values of a first output sensitivity according to each value of each input element of the second input element group in order for some values of a first intermediate variable to become zero, and the remaining values of the first intermediate variable to become discrete values other than zero;
set values of a first multiplex output sensitivity according to each value of each element of a multiplex input element group in order for some values of the first intermediate variable to become zero, and the remaining values of the first intermediate variable to become discrete values other than zero;
determine the first intermediate variable based upon a product of each value of input element of the first input element group, a value of the first output sensitivity, and the first multiplex output sensitivity such that the first intermediate variable is equal to zero when either the value of the first output sensitivity or the first multiplex output sensitivity is set equal to zero;
wherein the output element group of the multi-layered neural network comprises a plurality of output elements;
determine each output element based upon a value of output variables which is a product of said each value of intermediate element of the intermediate element group and a connection weight, wherein the values of output variables are zero or other than zero;
wherein the connection weight corresponds to an importance of values of the intermediate elements;
store correspondence relationship information that specifies a correspondence relationship between a plurality of classification information in order to classify an action of the subject, and a combination pattern of the values of said plurality of output elements which comprise discrete values of zero or discrete values other than zero;
estimate said classification information for the subject'"'"'s motion and information according to said first and second information based on said correspondence relationship information and the combination pattern of values of said plurality of output variables; and
display the classification information.
1 Assignment
0 Petitions
Accused Products
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.
-
Citations
9 Claims
-
1. A classification estimating system for a subject'"'"'s motion, based on a multi-layered neural network comprising:
-
a first measuring device configured to measure first information as a first biosignal according to a first measuring section of the subject; a second measuring device configured to measure second information as a second biosignal according to a second measuring section of the subject; a computer-readable memory storing executable instructions; and a processor in communication with the computer-readable memory and the first and second measuring devices, wherein the processor is programmed by the executable instructions to at least; use the multi-layered neural network, the multi-layered neural network comprising an input element group in an input layer of the multi-layered neural network, an intermediate element group in a hidden layer of the multi-layered neural network and an output element group in an output layer of the multi-layered neural network, for estimating an action of the subject; wherein the input element group of the multi-layered neural network comprises a first input element group and a second input element group, the first element group comprising a first plurality of input elements, the second element group comprising a second plurality of input elements; determine values of a first input variable based on the first information; provide each value of the first input variable to each of the first plurality of input elements of the first input element group; determine values of a second input variable based on the second information; provide each value of the second input variable to each of the second plurality of input elements of the second input element group; wherein the intermediate element group of the multi-layered neural network comprises a first intermediate element group, the first intermediate element group comprising a plurality of intermediate elements; set values of a first output sensitivity according to each value of each input element of the second input element group in order for some values of a first intermediate variable to become zero, and the remaining values of the first intermediate variable to become discrete values other than zero; set values of a first multiplex output sensitivity according to each value of each element of a multiplex input element group in order for some values of the first intermediate variable to become zero, and the remaining values of the first intermediate variable to become discrete values other than zero; determine the first intermediate variable based upon a product of each value of input element of the first input element group, a value of the first output sensitivity, and the first multiplex output sensitivity such that the first intermediate variable is equal to zero when either the value of the first output sensitivity or the first multiplex output sensitivity is set equal to zero; wherein the output element group of the multi-layered neural network comprises a plurality of output elements; determine each output element based upon a value of output variables which is a product of said each value of intermediate element of the intermediate element group and a connection weight, wherein the values of output variables are zero or other than zero; wherein the connection weight corresponds to an importance of values of the intermediate elements; store correspondence relationship information that specifies a correspondence relationship between a plurality of classification information in order to classify an action of the subject, and a combination pattern of the values of said plurality of output elements which comprise discrete values of zero or discrete values other than zero; estimate said classification information for the subject'"'"'s motion and information according to said first and second information based on said correspondence relationship information and the combination pattern of values of said plurality of output variables; and display the classification information. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8)
-
-
9. A classification estimating system for a subject'"'"'s motion comprising:
-
an input value memorizing unit adapted to memorize variables of a multi-layered neural network comprising an input element group in an input layer of the multi-layered neural network, an intermediate element group in an intermediate layer of the multi-layered neural network, and an output element group in an output layer of the multi-layered neural network; an intermediate value calculating unit adapted to calculate values of first intermediate variables based on input values stored in the input memorizing unit; an output calculating unit adapted to calculate values of output variables based on the first intermediate variables; a classification information estimating unit adapted to estimate classification information based on classified information of an activity of the subject, and further adapted to provide correspondence relationship information associated with a correspondence relationship between said value of output variables, and said value of output variables calculated by said value of output calculating unit; a classification information display unit adapted to display said classification information estimated according to said classification information estimating unit, wherein the input element group comprises a first input element group and a second input element group, the first element group comprising a plurality of first input elements, the second element group comprising a plurality of second input elements; a first measuring device adapted to measure first information as a biosignal according to a first measuring section of the subject; a processor programmed by executable instructions to; determine values of a first input variable based on the first information; provide each value of the first input variable to each first input element of the first input element group; a second measuring device adapted to measure second information as a biosignal according to a second measuring section of the subject; wherein the processor is further programmed by the executable instructions to; determine values of a second input variable based on the second information; provide each value of the second input variable to each second input element of the second input element group; wherein the intermediate element group comprises a first intermediate element group; wherein the first intermediate element group comprises a plurality of intermediate elements; wherein the processor is further programmed by the executable instructions to; set values of a first output sensitivity according to each value of input element of the second input element group in order for some values of a first intermediate variable to become zero, and the remaining values of the first intermediate variable to become discrete values other than zero; set values of a first multiplex output sensitivity according to each value of element of a pre-set multiplex input element group in order for some values of the first intermediate variable to become zero, and the remaining values of the first intermediate variable to become discrete values other than zero; and determine the first intermediate variable based upon a product of each value of input element of the first input element group, a value of the first output sensitivity, and the first multiplex output sensitivity such that the first intermediate variable is equal to zero when either the value of the first output sensitivity or the first multiplex output sensitivity is set equal to zero; wherein the output element group of the multi-layer neural network comprises a plurality of output elements; wherein each output element input by the value of output variables which is a product of said each value of intermediate element of the intermediate element group and a connection weight, wherein the values of output variables are zero or other than zero; wherein the connection weights correspond to importance of values of the intermediate elements; a correspondence relationship information memorizing unit adapted to memorize correspondence relationship information, which specifies a correspondence relationship between a plurality of classification information in order to classify an action of the subject, and the combination pattern of the value of said a plurality of output elements which comprise an output element group are zero or discrete values other than zero; wherein said classification information for the subject'"'"'s motion and information according to said first and second information estimated by the classification information estimating unit adapted to estimate based on said correspondence relationship information pre-stored in said correspondence relationship memorizing unit and the calculated combination pattern of values of said plurality of output variables are zero or discrete values other than zero.
-
Specification