Pattern recognition device using a neural network
AssigneeSharp Kabushiki Kaisha
InventorShin Kamiya
Life of this patent
Estimating the term…
Abstract
A pattern recognition device has a DP matching section. The DP matching section performs frequency expansion DP matching to a standard pattern and a characteristic pattern obtained from input voice waveform to obtain a DP score and DP path pattern. It is determined by means of a category identification neural network using the DP path pattern obtained from the DP matching section whether a category of the standard pattern and a category of the characteristic pattern are the same, and a determination result corresponding to the degree of identification is obtained. A normalized DP score, which is the DP score normalized for individual differences within a required range, is then obtained in a divider by compensating the DP score using the determination result.
Claims5
A pattern recognition device comprising:
- a dynamic programming matching section which applies a dynamic programming matching to both a characteristic pattern which expresses characteristics of input information and a standard pattern which expresses each of multiple categories inclusive of a category to which the input information belongs, and which thereby generates a dynamic programming score expressing a total sum of distances along an optimum path and a dynamic programming path pattern expressing the optimum path as a matrix pattern;
- a category identification neural network to which are input values of elements of the dynamic programming path pattern generated by the dynamic programming matching section, and which determines whether a category of the characteristic pattern and a category of the standard pattern are the same, and which outputs a determination result which expresses the degree of the determination;
- and
- a normalized dynamic programming score generator which compensates the dynamic programming score generated by the dynamic programming matching section using the determination result output from the category identification neural network, and thus generates a normalized dynamic programming score normalizing an allowable distortion in the characteristic pattern.
Dependencies are inferred from the claim text.