| 摘要 |
PURPOSE: To provide a method for classifying signals indicating each one of a plurality of possible amplitude values of a signal which can be matched easily to various kinds of roles and has learning processes which are made easier to the maximum. CONSTITUTION: Samples P1-P32 are respectively formed in parallel from a plurality of sampled values in one or a plurality of channels K1-K32 and at least one feature is extracted from the samples P1-P32. Then one or a plurality of features of the samples P1-P32 is used as addresses for reading out a stored probability of occurrence and, for succeeding evaluation, a discriminating amount is calculated from the probabilities of occurrence of all samples P1-P32 and compared with a threshold. |
