| 摘要 |
<p>PROBLEM TO BE SOLVED: To solve the following problems of a conventional correlation processor: when the presence or absence of correlation between data a and data b each having an error is determined, the determination is made based on distribution of information on errors of the data a and data b (information prior distribution), and therefore as an error without information increases, wrong determination of the presence or absence of correlation also increases; and an error of a calculated correlation probability also increases.SOLUTION: A correlation processor includes at least one of: a correlation determination part for determining the presence or absence of correlation between data a and data b, each having an error, on the basis of a probability density function of distribution of sum of distribution of information on the errors of the data a and data b (information prior distribution) and distribution without information (no-information prior distribution) or an approximate function of the probability density function; and a correlation probability calculation part for calculating a correlation probability between the data a and the data b, each having an error.</p> |
