摘要 |
<p>PURPOSE:To obtain a data searching method which can adaptively change the likelihood by optimizing the bias value used by a sequential coding means based on a standard data series. CONSTITUTION:A sequential decoder 104 reads out the probabilities of the data series held by a storage circuit 102 and the standard series. Then the decoder 104 input the read-out probabilities to a function generator 106 which calculates a logarithm of the value obtained by dividing the conditional probability of the partial series of the input data series calculated when the partial series of the standard data series is given by the probability of the partial series of the input data series and to a function generator 107 which calculates a logarithm of the probability of the partial series of the standard data series. Then the outputs of both generators 106 end 107 are added together by an adder 105 and a sequential decoding operation is carried out. As a result, the likelihood is automatically optimized and the data can be searched at a high speed even though the number of pattern recognizers and standard data series which are registered in an expert system, etc., are made numerous.</p> |