Analysis device, analysis method, and program

Abstract

An analysis device which analyzes a system that inputs input data including a plurality of input parameters and outputs output data, including an acquisition unit that acquires learning data including a plurality of sets of the input data and the output data, and a learning processing unit that learns, based on the acquired learning data, the amount of difference of output data corresponding to a difference between input parameters of two pieces of input data, an analysis method using the analysis device, and a program used in the analysis device are provided.

What is claimed is: 
     
         1 . An analysis device for analyzing a system that inputs input data including a plurality of input parameters and outputs output data, the device comprising:
 an acquisition unit that acquires learning data including a plurality of sets of the input data and the output data; and   a learning processing unit that learns, based on the acquired learning data, an amount of difference of output data corresponding to a difference between input parameters of two pieces of input data.   
     
     
         2 . The analysis device according to  claim 1 , wherein the learning processing unit generates an estimation model for learning a distance between two pieces of output data with respect to the difference between the input parameters of the two pieces of input data and estimating a change of the output data with respect to a change of the input parameters. 
     
     
         3 . The analysis device according to  claim 1 , wherein the learning processing unit performs pair-wise regression to perform regression analysis of the relationship between the difference between the input parameters and the amount of difference of the output data for each pair. 
     
     
         4 . The analysis device according to  claim 2 , wherein the learning processing unit generates an estimation model for estimating, by using a degree of change for every range of a value between the two input parameters, the amount of difference of the output data for the every range of the value between the input parameters. 
     
     
         5 . The analysis device according to  claim 2 , further comprising:
 an estimation unit that estimates, based on the estimation model, an amount of change of output data with respect to an amount of change of the input data.   
     
     
         6 . The analysis device according to  claim 1 , further comprising:
 a normalization unit that performs, for the plurality of pieces of input data, normalization of the input parameters so that an average of the input parameters is 0 and a variance of the input parameters is 1.   
     
     
         7 . The analysis device according to  claim 2 , further comprising:
 a display unit that displays, in accordance with an amount of change of the input parameters, an estimated amount of change of the output data.   
     
     
         8 . The analysis device according to  claim 1 , wherein the input parameters include an initial condition in a collision simulation, and wherein the output data includes shape data of an object in the collision simulation. 
     
     
         9 . An analysis method for analyzing a system that inputs input data including a plurality of input parameters and outputs output data, the method comprising:
 an acquisition step of acquiring learning data including a plurality of sets of the input data and the output data; and   a learning processing step of learning, based on the acquired learning data, an amount of difference of output data corresponding to a difference between input parameters of two pieces of input data.   
     
     
         10 . The analysis method according to  claim 9 , wherein the learning processing step includes generating an estimation model for learning a distance between two pieces of output data with respect to the difference between the input parameters of the two pieces of input data and estimating a change of the output data with respect to a change of the input parameters is generated. 
     
     
         11 . The analysis method according to  claim 10 , wherein the learning processing step includes performing pair-wise regression to perform regression analysis of the relationship between the difference between the input parameters and the amount of difference of the output data for each pair. 
     
     
         12 . The analysis method according to  claim 10 , wherein the learning processing step includes generating an estimation model for estimating, by using a degree of change for every range of a value between the two input parameters, the amount of difference of the output data for the every range of the value between the input parameters is generated. 
     
     
         13 . The analysis method according to  claim 10 , further comprising:
 an estimation step of estimating, based on the estimation model, an amount of change of output data with respect to an amount of change of the input data.   
     
     
         14 . The analysis method according to  claim 9 , further comprising:
 a normalization step of performing, for the plurality of pieces of input data, normalization of the input parameters so that an average of the input parameters is 0 and a variance of the input parameters is 1.   
     
     
         15 . The analysis method according to  claim 10 , further comprising:
 a display step of displaying, in accordance with an amount of change of the input parameters, an estimated amount of change of the output data.   
     
     
         16 . The analysis method according to  claim 9 , wherein the input parameters include an initial condition in a collision simulation, and wherein the output data includes shape data of an object in the collision simulation. 
     
     
         17 . A computer program product for analyzing a system that inputs input data including a plurality of input parameters and outputs output data, the computer program product comprising at least one computer readable non-transitory storage medium having computer readable program instructions thereon for execution by a processor, the computer readable program instructions comprising program instructions for:
 acquiring learning data including a plurality of sets of the input data and the output data; and   learning, based on the acquired learning data, an amount of difference of output data corresponding to a difference between input parameters of two pieces of input data.   
     
     
         18 . The computer program product according to  claim 17 , wherein the learning includes generating an estimation model for learning a distance between two pieces of output data with respect to the difference between the input parameters of the two pieces of input data and estimating a change of the output data with respect to a change of the input parameters is generated. 
     
     
         19 . The computer program product according to  claim 18 , wherein the learning includes performing pair-wise regression to perform regression analysis of the relationship between the difference between the input parameters and the amount of difference of the output data for each pair. 
     
     
         20 . The computer program product according to  claim 18 , wherein the learning includes generating an estimation model for estimating, by using a degree of change for every range of a value between the two input parameters, the amount of difference of the output data for the every range of the value between the input parameters is generated.