STATISTICAL DATA RECONSTRUCTION DEVICE, STATISTICAL DATA RECONSTRUCTION METHOD, AND PROGRAM

摘要

PROBLEM TO BE SOLVED: To provide a statistical data reconstruction device with which it is possible to reconstruct statistical data with high accuracy using a fewer number of data than conventional.SOLUTION: The statistical data reconstruction device includes: a parameter initialization unit which, assuming that the original data before disturbance processing can be described by a linear sum of a finite number of Gauss distributions of nondiagonal elements 0 of a covariance matrix, initializes the weight, average, and dispersion of each Gauss distribution; an E step calculation unit for repeatedly executing a process of calculating a burden rate for all combinations of a prescribed number of disturbance data generated by adding prescribed noise to the original data and Gauss distributions having a prescribed number of noises added thereto; an M step calculation unit for repeatedly executing a process of updating the weight, average, dispersion of each of the Gauss distributions; and a statistical data reconstruction unit for reconstructing statistical data using the converged weight, average, dispersion.SELECTED DRAWING: Figure 1