发明名称 DATA GENERATION PROBABILITY ESTIMATING DEVICE
摘要 PROBLEM TO BE SOLVED: To improve a method which is known so far so as to have minimum and maximum properties when relative redundancy is used for a general model class by mixing establishment density obtained by an establishment density calculating device at a very small rate, performing calculation based upon a specific expression, and outputting the result. SOLUTION: The establishment density calculating device 1 inputs a column xt and a vector parameterθand outputs the establishment density p(Xtθ) of xt. A Bayes mixed calculation device 2 calculates and outputs an approximate value of Bayes mixing density pw(xn) based upon a previous distribution w(θ) predetermined by using the establishment density calculating device. In addition to those devices, a device which calculates and outputs the approximate value of Bayes mixing density m(xn) of an index group mixture group handle by using the establishment density calculating device 1 and a device which calculates (1-ε)pw(xn)+ε.m(xn) by mixing m(xn) with the obtained establishment density pw(xn) at a slight rateεand outputs it are included.
申请公布号 JP2000163404(A) 申请公布日期 2000.06.16
申请号 JP19980334546 申请日期 1998.11.25
申请人 NEC CORP;REAL WORLD COMPUTING PARTNERSHIP 发明人 TAKEUCHI JUNICHI
分类号 G06F17/18;(IPC1-7):G06F17/18 主分类号 G06F17/18
代理机构 代理人
主权项
地址